diff --git a/.gitattributes b/.gitattributes
index a6344aac8c09253b3b630fb776ae94478aa0275b..ea9ca314c90130339ff41c9a12fd0660151f8288 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -33,3 +33,24 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-19200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-20000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-20800/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-21600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-22400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-23200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-24000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-24800/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-25600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-26400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-27200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-28000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-28800/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-29600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-30400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-31200/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-32000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-32800/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-33600/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-34400/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-epoch-4/tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/checkpoint-19200/README.md b/checkpoint-19200/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-19200/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-19200/adapter_config.json b/checkpoint-19200/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-19200/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-19200/adapter_model.safetensors b/checkpoint-19200/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..d18b78ccd4c70e58adef6a013ff8dbb1c6504df7
--- /dev/null
+++ b/checkpoint-19200/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:285fd80afeed0c673cd24ce99ab039194c8fa120f6a1db2001e02bf9571beb6d
+size 313820248
diff --git a/checkpoint-19200/rng_state.pth b/checkpoint-19200/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b02bdda2a338109ace1e988d557d0dd1809987ff
--- /dev/null
+++ b/checkpoint-19200/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e9c79e040cd1d90c9235f5ea004ce98bb1df30785570b9fdcf2a9e29948f896
+size 14244
diff --git a/checkpoint-19200/scheduler.pt b/checkpoint-19200/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..24892875cf178b396ba5af8a8a464ecb1c4dacd1
--- /dev/null
+++ b/checkpoint-19200/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3348413c5565e84f72b5de03d822ed1866796c840ac5265b9b52f983f212aba
+size 1064
diff --git a/checkpoint-19200/special_tokens_map.json b/checkpoint-19200/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-19200/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-19200/tokenizer.json b/checkpoint-19200/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-19200/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-19200/tokenizer.model b/checkpoint-19200/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-19200/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-19200/tokenizer_config.json b/checkpoint-19200/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-19200/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-19200/trainer_state.json b/checkpoint-19200/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ce687d4cd4135bcd294d26e41fe32be55144bdf2
--- /dev/null
+++ b/checkpoint-19200/trainer_state.json
@@ -0,0 +1,1941 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.187161815799966,
+  "eval_steps": 200,
+  "global_step": 19200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 1.9452726542216356e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-19200/training_args.bin b/checkpoint-19200/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-19200/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-20000/README.md b/checkpoint-20000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-20000/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-20000/adapter_config.json b/checkpoint-20000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-20000/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-20000/adapter_model.safetensors b/checkpoint-20000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..a43055dab062084091afb361d1e61923ab2b7b99
--- /dev/null
+++ b/checkpoint-20000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba64a48660d96e6ae210c02e7b193ff3ddad9dba223c1a63500a488145c438cb
+size 313820248
diff --git a/checkpoint-20000/rng_state.pth b/checkpoint-20000/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..5bea8cf270074197b814c88cc7b8eaf7a70101a4
--- /dev/null
+++ b/checkpoint-20000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fba149b3295ba204666aba65c69a1092c2683c7409ae96a94ad417bde27841b
+size 14244
diff --git a/checkpoint-20000/scheduler.pt b/checkpoint-20000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..9a511e0afe058eabc678ef484cf566b7cec6dbe8
--- /dev/null
+++ b/checkpoint-20000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de935c76c515679cf99a388f4ac63df38c3b321d9fb6d4b28cb9bf7338ecb1af
+size 1064
diff --git a/checkpoint-20000/special_tokens_map.json b/checkpoint-20000/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-20000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-20000/tokenizer.json b/checkpoint-20000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-20000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-20000/tokenizer.model b/checkpoint-20000/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-20000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-20000/tokenizer_config.json b/checkpoint-20000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-20000/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-20000/trainer_state.json b/checkpoint-20000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..2bf8afe1dfe31a24cea0ce87faac81137812ef1c
--- /dev/null
+++ b/checkpoint-20000/trainer_state.json
@@ -0,0 +1,2021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.2782935581249646,
+  "eval_steps": 200,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.0270434657179034e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-20000/training_args.bin b/checkpoint-20000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-20000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-20800/README.md b/checkpoint-20800/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-20800/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-20800/adapter_config.json b/checkpoint-20800/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-20800/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-20800/adapter_model.safetensors b/checkpoint-20800/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..b70c354071bc2d0946908436eb25209acd146a02
--- /dev/null
+++ b/checkpoint-20800/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8bbe4df1701bf265adab036c3e96cb82f9cb29bffaf953f7cc1cc5f43da0a96a
+size 313820248
diff --git a/checkpoint-20800/rng_state.pth b/checkpoint-20800/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2e21727ffbb22dd98361009a5ac826098e017aca
--- /dev/null
+++ b/checkpoint-20800/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:769c9d116d716759c38c18535106ad97215ee71bcb7213d14aa01a6e2eeb4af3
+size 14244
diff --git a/checkpoint-20800/scheduler.pt b/checkpoint-20800/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..d144dd8e4a1fd3ef9443a7d08442ba178e8ce74e
--- /dev/null
+++ b/checkpoint-20800/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c63a677b50085086e66ea55fbe01bc8715f3747ab6b695c5ef7d013aa1811797
+size 1064
diff --git a/checkpoint-20800/special_tokens_map.json b/checkpoint-20800/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-20800/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-20800/tokenizer.json b/checkpoint-20800/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-20800/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-20800/tokenizer.model b/checkpoint-20800/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-20800/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-20800/tokenizer_config.json b/checkpoint-20800/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-20800/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-20800/trainer_state.json b/checkpoint-20800/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..b976709d88d31d02c2463501b1dbc78375691b73
--- /dev/null
+++ b/checkpoint-20800/trainer_state.json
@@ -0,0 +1,2101 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.369425300449963,
+  "eval_steps": 200,
+  "global_step": 20800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.1077602489057567e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-20800/training_args.bin b/checkpoint-20800/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-20800/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-21600/README.md b/checkpoint-21600/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-21600/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-21600/adapter_config.json b/checkpoint-21600/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-21600/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-21600/adapter_model.safetensors b/checkpoint-21600/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..3c7d4d5b438ab15833ebef55d874d8eedde411c6
--- /dev/null
+++ b/checkpoint-21600/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a9fcd274c0981e09f7a56fc7df572e5f0b95bc58848f8bf7531f5054e18c99c
+size 313820248
diff --git a/checkpoint-21600/rng_state.pth b/checkpoint-21600/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..12dc1ecf71ebcf9a4a41a65cd6055b94b2259f9f
--- /dev/null
+++ b/checkpoint-21600/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a56b97f12072d267caf15b1f0d1101714ba6380a7c60c5a83e68808dc36d773
+size 14244
diff --git a/checkpoint-21600/scheduler.pt b/checkpoint-21600/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..0e33f99d70544ebf88d6064afd34f27b4e5f629e
--- /dev/null
+++ b/checkpoint-21600/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4ca451b22b6741c0b9a90d4a0705be6a2e71a2b099ab0a034a9a7bd72141ef9
+size 1064
diff --git a/checkpoint-21600/special_tokens_map.json b/checkpoint-21600/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-21600/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-21600/tokenizer.json b/checkpoint-21600/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-21600/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-21600/tokenizer.model b/checkpoint-21600/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-21600/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-21600/tokenizer_config.json b/checkpoint-21600/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-21600/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-21600/trainer_state.json b/checkpoint-21600/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..7a888360a5e810195905df7d412f13995097ef28
--- /dev/null
+++ b/checkpoint-21600/trainer_state.json
@@ -0,0 +1,2181 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4605570427749615,
+  "eval_steps": 200,
+  "global_step": 21600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.1880581377217454e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-21600/training_args.bin b/checkpoint-21600/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-21600/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-22400/README.md b/checkpoint-22400/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-22400/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-22400/adapter_config.json b/checkpoint-22400/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-22400/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-22400/adapter_model.safetensors b/checkpoint-22400/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..22675df9862808780c09a6401b5d0d63ff726904
--- /dev/null
+++ b/checkpoint-22400/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd6015a3bec35ad1dc604e30fb01a9c051ef9b3e36304464382ec41c027196bb
+size 313820248
diff --git a/checkpoint-22400/rng_state.pth b/checkpoint-22400/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8bb51d98c4c5c20fa950c2adf39fbe7f2cfa8add
--- /dev/null
+++ b/checkpoint-22400/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51466a5ddd7196f688cac7aee47b26b078df13365d4372200c01a591729462c5
+size 14244
diff --git a/checkpoint-22400/scheduler.pt b/checkpoint-22400/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..a60bd9228efb457e6464661699c10419c0b65262
--- /dev/null
+++ b/checkpoint-22400/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:512b506129b7457fd633b55ef740589dea9c8726485f78ed74b0523c9944ff7f
+size 1064
diff --git a/checkpoint-22400/special_tokens_map.json b/checkpoint-22400/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-22400/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-22400/tokenizer.json b/checkpoint-22400/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-22400/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-22400/tokenizer.model b/checkpoint-22400/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-22400/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-22400/tokenizer_config.json b/checkpoint-22400/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-22400/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-22400/trainer_state.json b/checkpoint-22400/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..c57fbd8dca5838f20424f297c559d59d8d94d45f
--- /dev/null
+++ b/checkpoint-22400/trainer_state.json
@@ -0,0 +1,2261 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.55168878509996,
+  "eval_steps": 200,
+  "global_step": 22400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.268984368095531e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-22400/training_args.bin b/checkpoint-22400/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-22400/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-23200/.ipynb_checkpoints/trainer_state-checkpoint.json b/checkpoint-23200/.ipynb_checkpoints/trainer_state-checkpoint.json
new file mode 100644
index 0000000000000000000000000000000000000000..df4423963e8d15d0d15471350bb998e125d2726f
--- /dev/null
+++ b/checkpoint-23200/.ipynb_checkpoints/trainer_state-checkpoint.json
@@ -0,0 +1,2341 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.6428205274249588,
+  "eval_steps": 200,
+  "global_step": 23200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.349948251199025e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-23200/README.md b/checkpoint-23200/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-23200/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-23200/adapter_config.json b/checkpoint-23200/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-23200/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-23200/adapter_model.safetensors b/checkpoint-23200/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fcc8f3ba75ce03d045ca7eba21c624a032cd4479
--- /dev/null
+++ b/checkpoint-23200/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26455c5cf186dd176cd767a5b9d3c00e63e8ca657920dada1897ddff98d2638e
+size 313820248
diff --git a/checkpoint-23200/rng_state.pth b/checkpoint-23200/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..23e1e7774a1a1013575b2ef8dbefea4031c873ab
--- /dev/null
+++ b/checkpoint-23200/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1a593379af6b00252592acd68b6ca2420c2a3f929054fac5a08b5ae47f6af27
+size 14244
diff --git a/checkpoint-23200/scheduler.pt b/checkpoint-23200/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..5f9528ed2116cf2326e4adcd917b1692dc9febeb
--- /dev/null
+++ b/checkpoint-23200/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9317957358d054605eac11acfffaf36b4209211ba63f1a7d8e2c7136e45a54b
+size 1064
diff --git a/checkpoint-23200/special_tokens_map.json b/checkpoint-23200/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-23200/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-23200/tokenizer.json b/checkpoint-23200/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-23200/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-23200/tokenizer.model b/checkpoint-23200/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-23200/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-23200/tokenizer_config.json b/checkpoint-23200/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-23200/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-23200/trainer_state.json b/checkpoint-23200/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..df4423963e8d15d0d15471350bb998e125d2726f
--- /dev/null
+++ b/checkpoint-23200/trainer_state.json
@@ -0,0 +1,2341 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.6428205274249588,
+  "eval_steps": 200,
+  "global_step": 23200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.349948251199025e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-23200/training_args.bin b/checkpoint-23200/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-23200/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-24000/.ipynb_checkpoints/trainer_state-checkpoint.json b/checkpoint-24000/.ipynb_checkpoints/trainer_state-checkpoint.json
new file mode 100644
index 0000000000000000000000000000000000000000..50f16476d3e3abab86500830765574b9955b165f
--- /dev/null
+++ b/checkpoint-24000/.ipynb_checkpoints/trainer_state-checkpoint.json
@@ -0,0 +1,2421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.7339522697499574,
+  "eval_steps": 200,
+  "global_step": 24000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.430740494923436e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-24000/README.md b/checkpoint-24000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-24000/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-24000/adapter_config.json b/checkpoint-24000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-24000/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-24000/adapter_model.safetensors b/checkpoint-24000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..6e3d3122d4df47c08174de9df08dcf6c13c303b6
--- /dev/null
+++ b/checkpoint-24000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a6d1df5f8c7cd3ede78d5212d9b192198cc15f6ae6ef071d8cebbb53b4540ab
+size 313820248
diff --git a/checkpoint-24000/rng_state.pth b/checkpoint-24000/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..6ab248464f18b3be456ef6e717ad34f4d89e0cdd
--- /dev/null
+++ b/checkpoint-24000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e54f05bd47371ea77b4ece351bf957d9d8ea03ad3a5f466a0c7cf4063093c7ae
+size 14244
diff --git a/checkpoint-24000/scheduler.pt b/checkpoint-24000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..41b73052261beaccd60d809719ba38870d144155
--- /dev/null
+++ b/checkpoint-24000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8148143e264b9a3313e0428f911ddffade2bff9601069f87e03b2f4e6928142d
+size 1064
diff --git a/checkpoint-24000/special_tokens_map.json b/checkpoint-24000/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-24000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-24000/tokenizer.json b/checkpoint-24000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-24000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-24000/tokenizer.model b/checkpoint-24000/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-24000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-24000/tokenizer_config.json b/checkpoint-24000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-24000/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-24000/trainer_state.json b/checkpoint-24000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..50f16476d3e3abab86500830765574b9955b165f
--- /dev/null
+++ b/checkpoint-24000/trainer_state.json
@@ -0,0 +1,2421 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.7339522697499574,
+  "eval_steps": 200,
+  "global_step": 24000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.430740494923436e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-24000/training_args.bin b/checkpoint-24000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-24000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-24800/README.md b/checkpoint-24800/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-24800/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-24800/adapter_config.json b/checkpoint-24800/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-24800/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-24800/adapter_model.safetensors b/checkpoint-24800/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..23824d1dfc030d911ce4809576ea2ab5583b917e
--- /dev/null
+++ b/checkpoint-24800/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13adac7c80b7f29760c1169e26168436cc9c0cff114938344758f1ec89be1e92
+size 313820248
diff --git a/checkpoint-24800/rng_state.pth b/checkpoint-24800/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d862bd9b9866981c90230313f8b34432f2196fbb
--- /dev/null
+++ b/checkpoint-24800/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d12a503836d4a5662c0b1cd380a7c60483c1e04c53a5e81895c52abaa71ba55d
+size 14244
diff --git a/checkpoint-24800/scheduler.pt b/checkpoint-24800/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..353e45926460c3d1573cbce883ce7fc3105c68d2
--- /dev/null
+++ b/checkpoint-24800/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:faccabdd4afc329266d1d679b1b8ed3659266a90b9838f1c7e3f1007f3403153
+size 1064
diff --git a/checkpoint-24800/special_tokens_map.json b/checkpoint-24800/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-24800/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-24800/tokenizer.json b/checkpoint-24800/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-24800/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-24800/tokenizer.model b/checkpoint-24800/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-24800/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-24800/tokenizer_config.json b/checkpoint-24800/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-24800/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-24800/trainer_state.json b/checkpoint-24800/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..414b4a610ef7789919af3a45a2b8865596ec30e1
--- /dev/null
+++ b/checkpoint-24800/trainer_state.json
@@ -0,0 +1,2501 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8250840120749556,
+  "eval_steps": 200,
+  "global_step": 24800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.511593466856059e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-24800/training_args.bin b/checkpoint-24800/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-24800/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-25600/README.md b/checkpoint-25600/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-25600/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-25600/adapter_config.json b/checkpoint-25600/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-25600/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-25600/adapter_model.safetensors b/checkpoint-25600/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..5f05c39adeabc4e6e1daec39dc6056667359ce86
--- /dev/null
+++ b/checkpoint-25600/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:babd2156bc62e1f21e027719a64eac31ba035890a56e2f9f458ac36a5e6ceff9
+size 313820248
diff --git a/checkpoint-25600/rng_state.pth b/checkpoint-25600/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..bcca3674321d3897169960e636181e5b9379bcc5
--- /dev/null
+++ b/checkpoint-25600/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be7cfde7487fd013eee571d563ee4399b8f447395e9428cd26c1055bda2e4e8d
+size 14244
diff --git a/checkpoint-25600/scheduler.pt b/checkpoint-25600/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..e3d55115c224470f333b1b9e8a73f5dc03bfbf49
--- /dev/null
+++ b/checkpoint-25600/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:035e2463face2dfe2bdb8b4d98ff3d7e199981d7f8b021bffe674c87a4ebbf3b
+size 1064
diff --git a/checkpoint-25600/special_tokens_map.json b/checkpoint-25600/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-25600/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-25600/tokenizer.json b/checkpoint-25600/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-25600/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-25600/tokenizer.model b/checkpoint-25600/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-25600/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-25600/tokenizer_config.json b/checkpoint-25600/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-25600/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-25600/trainer_state.json b/checkpoint-25600/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..8beaf26befa6e3fdb670bc57b5871b97710777ad
--- /dev/null
+++ b/checkpoint-25600/trainer_state.json
@@ -0,0 +1,2581 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9162157543999543,
+  "eval_steps": 200,
+  "global_step": 25600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.5927358127328174e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-25600/training_args.bin b/checkpoint-25600/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-25600/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-26400/README.md b/checkpoint-26400/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-26400/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-26400/adapter_config.json b/checkpoint-26400/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-26400/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-26400/adapter_model.safetensors b/checkpoint-26400/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..2a7adb5d54ff9de730da690604c430c2e5d3ab80
--- /dev/null
+++ b/checkpoint-26400/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb2b120e2dff5880501e9bb1ff6d6f8007cfe81007e82db5c15492f1c59f45cd
+size 313820248
diff --git a/checkpoint-26400/rng_state.pth b/checkpoint-26400/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..982d07c89f079c435c7986c047ef52945294d048
--- /dev/null
+++ b/checkpoint-26400/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4059ee1264e8f9bba695f8281a7c28f23b738ca5a8d2cf20d3d600ce8d3cada
+size 14244
diff --git a/checkpoint-26400/scheduler.pt b/checkpoint-26400/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1cbca66cd0afdcb53938f94351a8fcb623d8092e
--- /dev/null
+++ b/checkpoint-26400/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1cd6930c13cf35d7781e67eda8f37ac136fcc9748be23606f23d25e3cc33340
+size 1064
diff --git a/checkpoint-26400/special_tokens_map.json b/checkpoint-26400/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-26400/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-26400/tokenizer.json b/checkpoint-26400/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-26400/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-26400/tokenizer.model b/checkpoint-26400/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-26400/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-26400/tokenizer_config.json b/checkpoint-26400/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-26400/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-26400/trainer_state.json b/checkpoint-26400/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..6ea76c170bdc50f2cac538b775e694c383c78ef1
--- /dev/null
+++ b/checkpoint-26400/trainer_state.json
@@ -0,0 +1,2661 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.007347496724953,
+  "eval_steps": 200,
+  "global_step": 26400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.673475345737179e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-26400/training_args.bin b/checkpoint-26400/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-26400/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-27200/README.md b/checkpoint-27200/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-27200/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-27200/adapter_config.json b/checkpoint-27200/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-27200/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-27200/adapter_model.safetensors b/checkpoint-27200/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..111d3e94602f4c69570172cb20a6ed6106365321
--- /dev/null
+++ b/checkpoint-27200/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80d748696b155950575826a96214dd1565d8b62822c8e4ce7f24bea560de77c9
+size 313820248
diff --git a/checkpoint-27200/rng_state.pth b/checkpoint-27200/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..778812c00b086bcbf759eaa87e201d228e18b661
--- /dev/null
+++ b/checkpoint-27200/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38d90e82a5fd3a9f6036c6f294a4b353787d6d622bc246a3da281d27ced8423e
+size 14244
diff --git a/checkpoint-27200/scheduler.pt b/checkpoint-27200/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..88d26ac72be2fd32ad2e1c7f258e98a690e5cb06
--- /dev/null
+++ b/checkpoint-27200/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9ddc3cca15a5a836617360248a76eac5abaaa1ac50b51c6d5f00e25ee918e80
+size 1064
diff --git a/checkpoint-27200/special_tokens_map.json b/checkpoint-27200/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-27200/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-27200/tokenizer.json b/checkpoint-27200/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-27200/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-27200/tokenizer.model b/checkpoint-27200/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-27200/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-27200/tokenizer_config.json b/checkpoint-27200/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-27200/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-27200/trainer_state.json b/checkpoint-27200/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ec5f3282298d7bbdba00b5430f3c567402b94f7d
--- /dev/null
+++ b/checkpoint-27200/trainer_state.json
@@ -0,0 +1,2741 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0984792390499516,
+  "eval_steps": 200,
+  "global_step": 27200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.754944191025504e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-27200/training_args.bin b/checkpoint-27200/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-27200/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-28000/README.md b/checkpoint-28000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-28000/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-28000/adapter_config.json b/checkpoint-28000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-28000/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-28000/adapter_model.safetensors b/checkpoint-28000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..195a7f4761c58ba6aa976833bdeb2a3dbe579288
--- /dev/null
+++ b/checkpoint-28000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1311c5b8cc3edea6168c639ff6a0c33051069e1990326ac8ae0435da6fee60f8
+size 313820248
diff --git a/checkpoint-28000/rng_state.pth b/checkpoint-28000/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..714e44bb07f61b9124d8043b791dcfb998e123ae
--- /dev/null
+++ b/checkpoint-28000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ef0bb393f621ee177c72bfbd5c6d8d67d07eb3f818de506f0cfe882da8196bc
+size 14244
diff --git a/checkpoint-28000/scheduler.pt b/checkpoint-28000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..8862d7e44e36bd3de458c2a26b72e80ae2dc3dd6
--- /dev/null
+++ b/checkpoint-28000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9660b410f53aafd17c1d5c4026358405dece5b689c094a823df22d3272b785e7
+size 1064
diff --git a/checkpoint-28000/special_tokens_map.json b/checkpoint-28000/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-28000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-28000/tokenizer.json b/checkpoint-28000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-28000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-28000/tokenizer.model b/checkpoint-28000/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-28000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-28000/tokenizer_config.json b/checkpoint-28000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-28000/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-28000/trainer_state.json b/checkpoint-28000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ab709b03c7b36bc3827940f339b735eca2c21ec1
--- /dev/null
+++ b/checkpoint-28000/trainer_state.json
@@ -0,0 +1,2821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.1896109813749502,
+  "eval_steps": 200,
+  "global_step": 28000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.8365319494651494e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-28000/training_args.bin b/checkpoint-28000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-28000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-28800/README.md b/checkpoint-28800/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-28800/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-28800/adapter_config.json b/checkpoint-28800/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-28800/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-28800/adapter_model.safetensors b/checkpoint-28800/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..4c317c198352ab77c36ef03b672fa04c4ef6db95
--- /dev/null
+++ b/checkpoint-28800/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6d7cb76dc3be3d24f3b2e758f5e5d6f650bdc6a55d8b8dbb84bd38f21d6a520
+size 313820248
diff --git a/checkpoint-28800/rng_state.pth b/checkpoint-28800/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..1f1c8e8949d19a5440ffe64be774b518117a4c1d
--- /dev/null
+++ b/checkpoint-28800/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f511bbdaff45458064a38f7de8bff228bfd8d9bcc6d507349cd152b11207683
+size 14244
diff --git a/checkpoint-28800/scheduler.pt b/checkpoint-28800/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..eb5e8e85ff2c95cfe6d2c096e92712fac79b702c
--- /dev/null
+++ b/checkpoint-28800/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:805db1f0189158e01ff4fabb90fffe76486b4a4dfc81f444771010a6f68ea70d
+size 1064
diff --git a/checkpoint-28800/special_tokens_map.json b/checkpoint-28800/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-28800/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-28800/tokenizer.json b/checkpoint-28800/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-28800/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-28800/tokenizer.model b/checkpoint-28800/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-28800/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-28800/tokenizer_config.json b/checkpoint-28800/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-28800/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-28800/trainer_state.json b/checkpoint-28800/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c8340fc4b991158aa30573a3a68411d573734e9
--- /dev/null
+++ b/checkpoint-28800/trainer_state.json
@@ -0,0 +1,2901 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.280742723699949,
+  "eval_steps": 200,
+  "global_step": 28800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.9178562938739753e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-28800/training_args.bin b/checkpoint-28800/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-28800/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-29600/README.md b/checkpoint-29600/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-29600/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-29600/adapter_config.json b/checkpoint-29600/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-29600/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-29600/adapter_model.safetensors b/checkpoint-29600/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..043ae6fc393033bd47f9cab2a0337ae1939b062b
--- /dev/null
+++ b/checkpoint-29600/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42ec8f0c2c1306c416d1e7b0c0a8320bfa78ddbb32da10c731921719fd9c68e5
+size 313820248
diff --git a/checkpoint-29600/rng_state.pth b/checkpoint-29600/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..9caf3979f0b959dea3093ef9f2efd663df1d9949
--- /dev/null
+++ b/checkpoint-29600/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e034ec692772e1fd873c3cb9aaf23af4463b78e8173f374db89af2013d5ee84
+size 14244
diff --git a/checkpoint-29600/scheduler.pt b/checkpoint-29600/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..3d7d334b7ed7fcec97848c2c1d6437ec794dc30e
--- /dev/null
+++ b/checkpoint-29600/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee03ecce41e214e1eb53f62707cd3d264b4e010b6ff23f820a2899d805db644f
+size 1064
diff --git a/checkpoint-29600/special_tokens_map.json b/checkpoint-29600/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-29600/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-29600/tokenizer.json b/checkpoint-29600/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-29600/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-29600/tokenizer.model b/checkpoint-29600/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-29600/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-29600/tokenizer_config.json b/checkpoint-29600/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-29600/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-29600/trainer_state.json b/checkpoint-29600/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..316f449b52046df9103618d8a273955fd67a7d28
--- /dev/null
+++ b/checkpoint-29600/trainer_state.json
@@ -0,0 +1,2981 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.3718744660249476,
+  "eval_steps": 200,
+  "global_step": 29600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 2.998441354538705e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-29600/training_args.bin b/checkpoint-29600/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-29600/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-30400/README.md b/checkpoint-30400/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-30400/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-30400/adapter_config.json b/checkpoint-30400/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-30400/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-30400/adapter_model.safetensors b/checkpoint-30400/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..9f93b011022bdfb70cc1547b987219fce2d22dac
--- /dev/null
+++ b/checkpoint-30400/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45a0ca2b594ef2bd51d5936c3b4b24749c68fe33121739a6be84512714828f43
+size 313820248
diff --git a/checkpoint-30400/rng_state.pth b/checkpoint-30400/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..59114fb4a8786545830c5777b670ed06b3c86f26
--- /dev/null
+++ b/checkpoint-30400/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77ff1b17025d788a1a68b124fdeda486bb7d2dbc822e176272f02f7c8b90a4e5
+size 14244
diff --git a/checkpoint-30400/scheduler.pt b/checkpoint-30400/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1259b9dc0cba58050f111dfa28e8f6d047c724cd
--- /dev/null
+++ b/checkpoint-30400/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0982dabc045546826fcfc82228c5a0c794381278b8a5ac05e6895fe43ef14c65
+size 1064
diff --git a/checkpoint-30400/special_tokens_map.json b/checkpoint-30400/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-30400/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-30400/tokenizer.json b/checkpoint-30400/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-30400/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-30400/tokenizer.model b/checkpoint-30400/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-30400/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-30400/tokenizer_config.json b/checkpoint-30400/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-30400/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-30400/trainer_state.json b/checkpoint-30400/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..bce4f0d63072afbe598b1b9a387c29a3bd76cb20
--- /dev/null
+++ b/checkpoint-30400/trainer_state.json
@@ -0,0 +1,3061 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.4630062083499458,
+  "eval_steps": 200,
+  "global_step": 30400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.080477813176615e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-30400/training_args.bin b/checkpoint-30400/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-30400/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-31200/README.md b/checkpoint-31200/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-31200/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-31200/adapter_config.json b/checkpoint-31200/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-31200/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-31200/adapter_model.safetensors b/checkpoint-31200/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..aa1969875252bd5911041410afe5cd088f800de9
--- /dev/null
+++ b/checkpoint-31200/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c24f50527a8f410c9d623c957198fd23e035f439057ee11c95dccf20ba1cb0b
+size 313820248
diff --git a/checkpoint-31200/rng_state.pth b/checkpoint-31200/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..1f3c53264d9c460a0d8b579f45accf06836e6108
--- /dev/null
+++ b/checkpoint-31200/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c538eb16551a277728aa79d4d803f545916f4d510128cbf75376b6b0248da3d
+size 14244
diff --git a/checkpoint-31200/scheduler.pt b/checkpoint-31200/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..69984bdfab7ebb47ced9b01ab88332edc57e7027
--- /dev/null
+++ b/checkpoint-31200/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9f1fbdcbc40935ae24855d74bc0599565956b37a6b9495d86b31c997395a74d
+size 1064
diff --git a/checkpoint-31200/special_tokens_map.json b/checkpoint-31200/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-31200/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-31200/tokenizer.json b/checkpoint-31200/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-31200/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-31200/tokenizer.model b/checkpoint-31200/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-31200/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-31200/tokenizer_config.json b/checkpoint-31200/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-31200/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-31200/trainer_state.json b/checkpoint-31200/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..4422c56aff9d74cbae73780a07726b0714696ef7
--- /dev/null
+++ b/checkpoint-31200/trainer_state.json
@@ -0,0 +1,3141 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.5541379506749444,
+  "eval_steps": 200,
+  "global_step": 31200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.1608547913342484e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-31200/training_args.bin b/checkpoint-31200/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-31200/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-32000/.ipynb_checkpoints/trainer_state-checkpoint.json b/checkpoint-32000/.ipynb_checkpoints/trainer_state-checkpoint.json
new file mode 100644
index 0000000000000000000000000000000000000000..8ec8925edf4f47c358896abd9394f7ec7a62d783
--- /dev/null
+++ b/checkpoint-32000/.ipynb_checkpoints/trainer_state-checkpoint.json
@@ -0,0 +1,3221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.645269692999943,
+  "eval_steps": 200,
+  "global_step": 32000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.240868671875162e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-32000/README.md b/checkpoint-32000/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-32000/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-32000/adapter_config.json b/checkpoint-32000/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-32000/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-32000/adapter_model.safetensors b/checkpoint-32000/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..fc5149319a1d53f160239462b233b0c330cd1ee1
--- /dev/null
+++ b/checkpoint-32000/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:842f81501ea50d9c7ac456b36ac301caebe2e984d9109a47ae42449ad6d3fb7d
+size 313820248
diff --git a/checkpoint-32000/rng_state.pth b/checkpoint-32000/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..e8221d6acffc7b20a1ce5f28b536a3311759975d
--- /dev/null
+++ b/checkpoint-32000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b6542aff70eaa7515635c81f34098614624ee88f307071dff5805996e8189f4
+size 14244
diff --git a/checkpoint-32000/scheduler.pt b/checkpoint-32000/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..1bb5ce192110dbf2cea1a7a233f173b3ac5aef0b
--- /dev/null
+++ b/checkpoint-32000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc598baa0a21eb1cd84e24ac1c8970564d4a747d342d0e671e37ec61b2b79412
+size 1064
diff --git a/checkpoint-32000/special_tokens_map.json b/checkpoint-32000/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-32000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-32000/tokenizer.json b/checkpoint-32000/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-32000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-32000/tokenizer.model b/checkpoint-32000/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-32000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-32000/tokenizer_config.json b/checkpoint-32000/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-32000/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-32000/trainer_state.json b/checkpoint-32000/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..8ec8925edf4f47c358896abd9394f7ec7a62d783
--- /dev/null
+++ b/checkpoint-32000/trainer_state.json
@@ -0,0 +1,3221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.645269692999943,
+  "eval_steps": 200,
+  "global_step": 32000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.240868671875162e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-32000/training_args.bin b/checkpoint-32000/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-32000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-32800/README.md b/checkpoint-32800/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-32800/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-32800/adapter_config.json b/checkpoint-32800/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-32800/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-32800/adapter_model.safetensors b/checkpoint-32800/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..e750e1ab7ca7532951fc3ae68fddf5078168de1e
--- /dev/null
+++ b/checkpoint-32800/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a6397dff9ec6176fa71f5a1c8cd2d66942c2592846de3ac2fb20ccbc98920eb
+size 313820248
diff --git a/checkpoint-32800/rng_state.pth b/checkpoint-32800/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..74a7525227518f48efe859997c56176ac106cd9c
--- /dev/null
+++ b/checkpoint-32800/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20874007e6b3e5b118e94ef14755cf600348e31812e64a025425d965057d7395
+size 14244
diff --git a/checkpoint-32800/scheduler.pt b/checkpoint-32800/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..2002d19505a14465c77f4501d54c19fdf3265c58
--- /dev/null
+++ b/checkpoint-32800/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f91aedfe935679b1a5eb71d383969820bfbf25516eef47b1ec627c1649050de3
+size 1064
diff --git a/checkpoint-32800/special_tokens_map.json b/checkpoint-32800/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-32800/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-32800/tokenizer.json b/checkpoint-32800/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-32800/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-32800/tokenizer.model b/checkpoint-32800/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-32800/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-32800/tokenizer_config.json b/checkpoint-32800/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-32800/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-32800/trainer_state.json b/checkpoint-32800/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..07639540e21fc66125488d9e4dc4d1de9ab2542d
--- /dev/null
+++ b/checkpoint-32800/trainer_state.json
@@ -0,0 +1,3301 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.7364014353249417,
+  "eval_steps": 200,
+  "global_step": 32800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.45725110173225403,
+      "learning_rate": 1.6589284188338508e-05,
+      "loss": 1.3042,
+      "step": 32200
+    },
+    {
+      "epoch": 3.67,
+      "eval_bertscore": 0.7283689379692078,
+      "eval_loss": 1.8278729915618896,
+      "eval_rouge1": 0.6438134996619729,
+      "eval_rouge2": 0.34631038919793855,
+      "eval_rougeL": 0.5185962520496568,
+      "eval_rougeLsum": 0.6335181171392547,
+      "eval_runtime": 14.9577,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 32200
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.527599036693573,
+      "learning_rate": 1.544991027430427e-05,
+      "loss": 1.3141,
+      "step": 32400
+    },
+    {
+      "epoch": 3.69,
+      "eval_bertscore": 0.7283642292022705,
+      "eval_loss": 1.828185796737671,
+      "eval_rouge1": 0.6420673962616871,
+      "eval_rouge2": 0.3428100609192179,
+      "eval_rougeL": 0.5151424402980586,
+      "eval_rougeLsum": 0.6312247707173357,
+      "eval_runtime": 15.0727,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 32400
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.4286324679851532,
+      "learning_rate": 1.4310536360270033e-05,
+      "loss": 1.2965,
+      "step": 32600
+    },
+    {
+      "epoch": 3.71,
+      "eval_bertscore": 0.7269189953804016,
+      "eval_loss": 1.827340841293335,
+      "eval_rouge1": 0.6423441986765093,
+      "eval_rouge2": 0.3444233896929453,
+      "eval_rougeL": 0.515230719711361,
+      "eval_rougeLsum": 0.6324502593091261,
+      "eval_runtime": 15.762,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.317,
+      "step": 32600
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.45857399702072144,
+      "learning_rate": 1.3171162446235794e-05,
+      "loss": 1.3106,
+      "step": 32800
+    },
+    {
+      "epoch": 3.74,
+      "eval_bertscore": 0.7266061305999756,
+      "eval_loss": 1.8283151388168335,
+      "eval_rouge1": 0.6411621669658236,
+      "eval_rouge2": 0.3443038317384875,
+      "eval_rougeL": 0.5144500926275494,
+      "eval_rougeLsum": 0.6309220735111227,
+      "eval_runtime": 15.7907,
+      "eval_samples_per_second": 0.633,
+      "eval_steps_per_second": 0.317,
+      "step": 32800
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.3218338576266363e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-32800/training_args.bin b/checkpoint-32800/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-32800/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-33600/.ipynb_checkpoints/trainer_state-checkpoint.json b/checkpoint-33600/.ipynb_checkpoints/trainer_state-checkpoint.json
new file mode 100644
index 0000000000000000000000000000000000000000..ce283174aebd90e995636ba95ac53df9d6c92a82
--- /dev/null
+++ b/checkpoint-33600/.ipynb_checkpoints/trainer_state-checkpoint.json
@@ -0,0 +1,3381 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.82753317764994,
+  "eval_steps": 200,
+  "global_step": 33600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.45725110173225403,
+      "learning_rate": 1.6589284188338508e-05,
+      "loss": 1.3042,
+      "step": 32200
+    },
+    {
+      "epoch": 3.67,
+      "eval_bertscore": 0.7283689379692078,
+      "eval_loss": 1.8278729915618896,
+      "eval_rouge1": 0.6438134996619729,
+      "eval_rouge2": 0.34631038919793855,
+      "eval_rougeL": 0.5185962520496568,
+      "eval_rougeLsum": 0.6335181171392547,
+      "eval_runtime": 14.9577,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 32200
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.527599036693573,
+      "learning_rate": 1.544991027430427e-05,
+      "loss": 1.3141,
+      "step": 32400
+    },
+    {
+      "epoch": 3.69,
+      "eval_bertscore": 0.7283642292022705,
+      "eval_loss": 1.828185796737671,
+      "eval_rouge1": 0.6420673962616871,
+      "eval_rouge2": 0.3428100609192179,
+      "eval_rougeL": 0.5151424402980586,
+      "eval_rougeLsum": 0.6312247707173357,
+      "eval_runtime": 15.0727,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 32400
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.4286324679851532,
+      "learning_rate": 1.4310536360270033e-05,
+      "loss": 1.2965,
+      "step": 32600
+    },
+    {
+      "epoch": 3.71,
+      "eval_bertscore": 0.7269189953804016,
+      "eval_loss": 1.827340841293335,
+      "eval_rouge1": 0.6423441986765093,
+      "eval_rouge2": 0.3444233896929453,
+      "eval_rougeL": 0.515230719711361,
+      "eval_rougeLsum": 0.6324502593091261,
+      "eval_runtime": 15.762,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.317,
+      "step": 32600
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.45857399702072144,
+      "learning_rate": 1.3171162446235794e-05,
+      "loss": 1.3106,
+      "step": 32800
+    },
+    {
+      "epoch": 3.74,
+      "eval_bertscore": 0.7266061305999756,
+      "eval_loss": 1.8283151388168335,
+      "eval_rouge1": 0.6411621669658236,
+      "eval_rouge2": 0.3443038317384875,
+      "eval_rougeL": 0.5144500926275494,
+      "eval_rougeLsum": 0.6309220735111227,
+      "eval_runtime": 15.7907,
+      "eval_samples_per_second": 0.633,
+      "eval_steps_per_second": 0.317,
+      "step": 32800
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.5014874935150146,
+      "learning_rate": 1.2031788532201556e-05,
+      "loss": 1.2949,
+      "step": 33000
+    },
+    {
+      "epoch": 3.76,
+      "eval_bertscore": 0.7257188558578491,
+      "eval_loss": 1.8288466930389404,
+      "eval_rouge1": 0.6423913963160501,
+      "eval_rouge2": 0.3441299434811822,
+      "eval_rougeL": 0.5144996687729989,
+      "eval_rougeLsum": 0.6320056022872689,
+      "eval_runtime": 15.28,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 33000
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.45499473810195923,
+      "learning_rate": 1.0892414618167317e-05,
+      "loss": 1.3217,
+      "step": 33200
+    },
+    {
+      "epoch": 3.78,
+      "eval_bertscore": 0.7278145551681519,
+      "eval_loss": 1.8302255868911743,
+      "eval_rouge1": 0.6425439428330799,
+      "eval_rouge2": 0.34625777040537886,
+      "eval_rougeL": 0.5154002067201906,
+      "eval_rougeLsum": 0.6311754713033721,
+      "eval_runtime": 14.9234,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 33200
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.528462827205658,
+      "learning_rate": 9.753040704133079e-06,
+      "loss": 1.2944,
+      "step": 33400
+    },
+    {
+      "epoch": 3.8,
+      "eval_bertscore": 0.7256959676742554,
+      "eval_loss": 1.8285919427871704,
+      "eval_rouge1": 0.6427339109412973,
+      "eval_rouge2": 0.34490192490238136,
+      "eval_rougeL": 0.5158098895481885,
+      "eval_rougeLsum": 0.6318354230613757,
+      "eval_runtime": 14.821,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 33400
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.4865368604660034,
+      "learning_rate": 8.61366679009884e-06,
+      "loss": 1.2974,
+      "step": 33600
+    },
+    {
+      "epoch": 3.83,
+      "eval_bertscore": 0.726431667804718,
+      "eval_loss": 1.827730417251587,
+      "eval_rouge1": 0.6411658367235875,
+      "eval_rouge2": 0.34535262367337394,
+      "eval_rougeL": 0.5145257214263663,
+      "eval_rougeLsum": 0.630130047089762,
+      "eval_runtime": 15.0627,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 33600
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.403299291217871e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-33600/README.md b/checkpoint-33600/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-33600/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-33600/adapter_config.json b/checkpoint-33600/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-33600/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-33600/adapter_model.safetensors b/checkpoint-33600/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..40210361fd9884882cf3f48d64af88c8c87c4161
--- /dev/null
+++ b/checkpoint-33600/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:102914b9bc2015fca57fd4855876582cd375028e4981ea663f52f269943db00c
+size 313820248
diff --git a/checkpoint-33600/rng_state.pth b/checkpoint-33600/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b69af059d50525ed67c4ecc930a6936abdef7dea
--- /dev/null
+++ b/checkpoint-33600/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0928eb54749672b323872e8e74fc514d9563ad7e41b0f9ade5715738846025c0
+size 14244
diff --git a/checkpoint-33600/scheduler.pt b/checkpoint-33600/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..288b324e030f1eecff83dbb29907db1647844e4c
--- /dev/null
+++ b/checkpoint-33600/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a370402239a91120272151cf81d8eaac76e01c295fca284c0b6273a3cb5e404b
+size 1064
diff --git a/checkpoint-33600/special_tokens_map.json b/checkpoint-33600/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-33600/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-33600/tokenizer.json b/checkpoint-33600/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-33600/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-33600/tokenizer.model b/checkpoint-33600/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-33600/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-33600/tokenizer_config.json b/checkpoint-33600/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-33600/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-33600/trainer_state.json b/checkpoint-33600/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..ce283174aebd90e995636ba95ac53df9d6c92a82
--- /dev/null
+++ b/checkpoint-33600/trainer_state.json
@@ -0,0 +1,3381 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.82753317764994,
+  "eval_steps": 200,
+  "global_step": 33600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.45725110173225403,
+      "learning_rate": 1.6589284188338508e-05,
+      "loss": 1.3042,
+      "step": 32200
+    },
+    {
+      "epoch": 3.67,
+      "eval_bertscore": 0.7283689379692078,
+      "eval_loss": 1.8278729915618896,
+      "eval_rouge1": 0.6438134996619729,
+      "eval_rouge2": 0.34631038919793855,
+      "eval_rougeL": 0.5185962520496568,
+      "eval_rougeLsum": 0.6335181171392547,
+      "eval_runtime": 14.9577,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 32200
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.527599036693573,
+      "learning_rate": 1.544991027430427e-05,
+      "loss": 1.3141,
+      "step": 32400
+    },
+    {
+      "epoch": 3.69,
+      "eval_bertscore": 0.7283642292022705,
+      "eval_loss": 1.828185796737671,
+      "eval_rouge1": 0.6420673962616871,
+      "eval_rouge2": 0.3428100609192179,
+      "eval_rougeL": 0.5151424402980586,
+      "eval_rougeLsum": 0.6312247707173357,
+      "eval_runtime": 15.0727,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 32400
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.4286324679851532,
+      "learning_rate": 1.4310536360270033e-05,
+      "loss": 1.2965,
+      "step": 32600
+    },
+    {
+      "epoch": 3.71,
+      "eval_bertscore": 0.7269189953804016,
+      "eval_loss": 1.827340841293335,
+      "eval_rouge1": 0.6423441986765093,
+      "eval_rouge2": 0.3444233896929453,
+      "eval_rougeL": 0.515230719711361,
+      "eval_rougeLsum": 0.6324502593091261,
+      "eval_runtime": 15.762,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.317,
+      "step": 32600
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.45857399702072144,
+      "learning_rate": 1.3171162446235794e-05,
+      "loss": 1.3106,
+      "step": 32800
+    },
+    {
+      "epoch": 3.74,
+      "eval_bertscore": 0.7266061305999756,
+      "eval_loss": 1.8283151388168335,
+      "eval_rouge1": 0.6411621669658236,
+      "eval_rouge2": 0.3443038317384875,
+      "eval_rougeL": 0.5144500926275494,
+      "eval_rougeLsum": 0.6309220735111227,
+      "eval_runtime": 15.7907,
+      "eval_samples_per_second": 0.633,
+      "eval_steps_per_second": 0.317,
+      "step": 32800
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.5014874935150146,
+      "learning_rate": 1.2031788532201556e-05,
+      "loss": 1.2949,
+      "step": 33000
+    },
+    {
+      "epoch": 3.76,
+      "eval_bertscore": 0.7257188558578491,
+      "eval_loss": 1.8288466930389404,
+      "eval_rouge1": 0.6423913963160501,
+      "eval_rouge2": 0.3441299434811822,
+      "eval_rougeL": 0.5144996687729989,
+      "eval_rougeLsum": 0.6320056022872689,
+      "eval_runtime": 15.28,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 33000
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.45499473810195923,
+      "learning_rate": 1.0892414618167317e-05,
+      "loss": 1.3217,
+      "step": 33200
+    },
+    {
+      "epoch": 3.78,
+      "eval_bertscore": 0.7278145551681519,
+      "eval_loss": 1.8302255868911743,
+      "eval_rouge1": 0.6425439428330799,
+      "eval_rouge2": 0.34625777040537886,
+      "eval_rougeL": 0.5154002067201906,
+      "eval_rougeLsum": 0.6311754713033721,
+      "eval_runtime": 14.9234,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 33200
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.528462827205658,
+      "learning_rate": 9.753040704133079e-06,
+      "loss": 1.2944,
+      "step": 33400
+    },
+    {
+      "epoch": 3.8,
+      "eval_bertscore": 0.7256959676742554,
+      "eval_loss": 1.8285919427871704,
+      "eval_rouge1": 0.6427339109412973,
+      "eval_rouge2": 0.34490192490238136,
+      "eval_rougeL": 0.5158098895481885,
+      "eval_rougeLsum": 0.6318354230613757,
+      "eval_runtime": 14.821,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 33400
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.4865368604660034,
+      "learning_rate": 8.61366679009884e-06,
+      "loss": 1.2974,
+      "step": 33600
+    },
+    {
+      "epoch": 3.83,
+      "eval_bertscore": 0.726431667804718,
+      "eval_loss": 1.827730417251587,
+      "eval_rouge1": 0.6411658367235875,
+      "eval_rouge2": 0.34535262367337394,
+      "eval_rougeL": 0.5145257214263663,
+      "eval_rougeLsum": 0.630130047089762,
+      "eval_runtime": 15.0627,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 33600
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.403299291217871e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-33600/training_args.bin b/checkpoint-33600/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-33600/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-34400/README.md b/checkpoint-34400/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-34400/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-34400/adapter_config.json b/checkpoint-34400/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-34400/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-34400/adapter_model.safetensors b/checkpoint-34400/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..1992b38fe0430f0003f32ba71f84d7fbd742d508
--- /dev/null
+++ b/checkpoint-34400/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dd5559abbed4295567c87430204fc1af2aaf488c1e45ab2931b05f80e802796
+size 313820248
diff --git a/checkpoint-34400/rng_state.pth b/checkpoint-34400/rng_state.pth
new file mode 100644
index 0000000000000000000000000000000000000000..72ea58363bf35a94425e41c61caeea97d60584d6
--- /dev/null
+++ b/checkpoint-34400/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da705518ae50c1e4a1c2c34db27f5f1831b96582da80eec17a4d5cb6549e1be3
+size 14244
diff --git a/checkpoint-34400/scheduler.pt b/checkpoint-34400/scheduler.pt
new file mode 100644
index 0000000000000000000000000000000000000000..313d1946e19b5ace7c6acb8f44800bef35e93ebb
--- /dev/null
+++ b/checkpoint-34400/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e7252842e011d1153efc0444ec8268ae2e16485e0347b6568dc34c66edb7638
+size 1064
diff --git a/checkpoint-34400/special_tokens_map.json b/checkpoint-34400/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-34400/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-34400/tokenizer.json b/checkpoint-34400/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-34400/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-34400/tokenizer.model b/checkpoint-34400/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-34400/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-34400/tokenizer_config.json b/checkpoint-34400/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-34400/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-34400/trainer_state.json b/checkpoint-34400/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..72093d9ecd1bd807ef3f16d898a4e297f4279464
--- /dev/null
+++ b/checkpoint-34400/trainer_state.json
@@ -0,0 +1,3461 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9186649199749386,
+  "eval_steps": 200,
+  "global_step": 34400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.1643640249967575,
+      "learning_rate": 0.0001988891104338166,
+      "loss": 1.7673,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_bertscore": 0.7312520742416382,
+      "eval_loss": 1.7944419384002686,
+      "eval_rouge1": 0.645726048132668,
+      "eval_rouge2": 0.342840307585653,
+      "eval_rougeL": 0.5174784271125388,
+      "eval_rougeLsum": 0.6359911842715976,
+      "eval_runtime": 67.7968,
+      "eval_samples_per_second": 0.147,
+      "eval_steps_per_second": 0.074,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.17238478362560272,
+      "learning_rate": 0.00019774973651978238,
+      "loss": 1.6985,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "eval_bertscore": 0.733666718006134,
+      "eval_loss": 1.7791178226470947,
+      "eval_rouge1": 0.6540909153028596,
+      "eval_rouge2": 0.3548819059818129,
+      "eval_rougeL": 0.527257232694246,
+      "eval_rougeLsum": 0.6442799950994005,
+      "eval_runtime": 15.1267,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.19368696212768555,
+      "learning_rate": 0.00019661036260574814,
+      "loss": 1.6962,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "eval_bertscore": 0.7339462041854858,
+      "eval_loss": 1.7609882354736328,
+      "eval_rouge1": 0.6384337329686338,
+      "eval_rouge2": 0.3415514270662107,
+      "eval_rougeL": 0.51206080148464,
+      "eval_rougeLsum": 0.6261968614666548,
+      "eval_runtime": 15.2068,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.18629203736782074,
+      "learning_rate": 0.00019547098869171392,
+      "loss": 1.6825,
+      "step": 800
+    },
+    {
+      "epoch": 0.09,
+      "eval_bertscore": 0.7363594174385071,
+      "eval_loss": 1.7610784769058228,
+      "eval_rouge1": 0.6461624591922237,
+      "eval_rouge2": 0.3477371388439609,
+      "eval_rougeL": 0.5187429174752844,
+      "eval_rougeLsum": 0.6361089823008282,
+      "eval_runtime": 15.173,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.1799013316631317,
+      "learning_rate": 0.00019433161477767967,
+      "loss": 1.6848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bertscore": 0.7334067225456238,
+      "eval_loss": 1.7576347589492798,
+      "eval_rouge1": 0.6345119236349537,
+      "eval_rouge2": 0.3422519149071803,
+      "eval_rougeL": 0.5111983101326238,
+      "eval_rougeLsum": 0.6244653120436832,
+      "eval_runtime": 15.2847,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.22036150097846985,
+      "learning_rate": 0.00019319224086364545,
+      "loss": 1.6714,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "eval_bertscore": 0.7323788404464722,
+      "eval_loss": 1.7521806955337524,
+      "eval_rouge1": 0.6452540184557478,
+      "eval_rouge2": 0.3465145726476423,
+      "eval_rougeL": 0.516711757588783,
+      "eval_rougeLsum": 0.6341049885677059,
+      "eval_runtime": 15.1247,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.21381086111068726,
+      "learning_rate": 0.0001920528669496112,
+      "loss": 1.6669,
+      "step": 1400
+    },
+    {
+      "epoch": 0.16,
+      "eval_bertscore": 0.7313202619552612,
+      "eval_loss": 1.7520482540130615,
+      "eval_rouge1": 0.6397526546254797,
+      "eval_rouge2": 0.3452671288110514,
+      "eval_rougeL": 0.5176580626678706,
+      "eval_rougeLsum": 0.6296746647539768,
+      "eval_runtime": 15.183,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.20332874357700348,
+      "learning_rate": 0.00019091349303557696,
+      "loss": 1.671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.18,
+      "eval_bertscore": 0.7349230647087097,
+      "eval_loss": 1.7473630905151367,
+      "eval_rouge1": 0.637439872504459,
+      "eval_rouge2": 0.34307164454056094,
+      "eval_rougeL": 0.5129717676228565,
+      "eval_rougeLsum": 0.6272190896182391,
+      "eval_runtime": 15.5672,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 1600
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2025599479675293,
+      "learning_rate": 0.00018977411912154274,
+      "loss": 1.6721,
+      "step": 1800
+    },
+    {
+      "epoch": 0.21,
+      "eval_bertscore": 0.7357184290885925,
+      "eval_loss": 1.7516342401504517,
+      "eval_rouge1": 0.6387615819926658,
+      "eval_rouge2": 0.34366787517105574,
+      "eval_rougeL": 0.5129026911770751,
+      "eval_rougeLsum": 0.6289314118258257,
+      "eval_runtime": 15.9574,
+      "eval_samples_per_second": 0.627,
+      "eval_steps_per_second": 0.313,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.20457112789154053,
+      "learning_rate": 0.0001886347452075085,
+      "loss": 1.671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.23,
+      "eval_bertscore": 0.733718752861023,
+      "eval_loss": 1.7501707077026367,
+      "eval_rouge1": 0.6346207681220664,
+      "eval_rouge2": 0.33748369437614106,
+      "eval_rougeL": 0.5085159047705141,
+      "eval_rougeLsum": 0.6239953154441167,
+      "eval_runtime": 15.0863,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.22552740573883057,
+      "learning_rate": 0.00018749537129347424,
+      "loss": 1.6496,
+      "step": 2200
+    },
+    {
+      "epoch": 0.25,
+      "eval_bertscore": 0.7368552684783936,
+      "eval_loss": 1.7437107563018799,
+      "eval_rouge1": 0.6490756387878311,
+      "eval_rouge2": 0.3448817738175175,
+      "eval_rougeL": 0.5235187045706321,
+      "eval_rougeLsum": 0.6377780857890332,
+      "eval_runtime": 15.07,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22573673725128174,
+      "learning_rate": 0.00018635599737944,
+      "loss": 1.6629,
+      "step": 2400
+    },
+    {
+      "epoch": 0.27,
+      "eval_bertscore": 0.7314499616622925,
+      "eval_loss": 1.7462828159332275,
+      "eval_rouge1": 0.6511482369678803,
+      "eval_rouge2": 0.34632544827771805,
+      "eval_rougeL": 0.5212417191003778,
+      "eval_rougeLsum": 0.6415391907940229,
+      "eval_runtime": 15.2078,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.26426687836647034,
+      "learning_rate": 0.00018521662346540575,
+      "loss": 1.6644,
+      "step": 2600
+    },
+    {
+      "epoch": 0.3,
+      "eval_bertscore": 0.7363359928131104,
+      "eval_loss": 1.7505037784576416,
+      "eval_rouge1": 0.6498296552335481,
+      "eval_rouge2": 0.34873833589761183,
+      "eval_rougeL": 0.5194028620820592,
+      "eval_rougeLsum": 0.6404603087578984,
+      "eval_runtime": 14.8403,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.20142091810703278,
+      "learning_rate": 0.00018407724955137153,
+      "loss": 1.6535,
+      "step": 2800
+    },
+    {
+      "epoch": 0.32,
+      "eval_bertscore": 0.7304679155349731,
+      "eval_loss": 1.7511039972305298,
+      "eval_rouge1": 0.6475130585388738,
+      "eval_rouge2": 0.34648331046897884,
+      "eval_rougeL": 0.5218042284020985,
+      "eval_rougeLsum": 0.6382749834402862,
+      "eval_runtime": 15.0162,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23283220827579498,
+      "learning_rate": 0.00018293787563733728,
+      "loss": 1.6477,
+      "step": 3000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bertscore": 0.7327049374580383,
+      "eval_loss": 1.7461665868759155,
+      "eval_rouge1": 0.6309349586871908,
+      "eval_rouge2": 0.3387882478990309,
+      "eval_rougeL": 0.5042059192403674,
+      "eval_rougeLsum": 0.6210432469674847,
+      "eval_runtime": 15.463,
+      "eval_samples_per_second": 0.647,
+      "eval_steps_per_second": 0.323,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.21316750347614288,
+      "learning_rate": 0.00018179850172330306,
+      "loss": 1.6614,
+      "step": 3200
+    },
+    {
+      "epoch": 0.36,
+      "eval_bertscore": 0.7314620018005371,
+      "eval_loss": 1.7468239068984985,
+      "eval_rouge1": 0.6480904534152265,
+      "eval_rouge2": 0.3479530168963481,
+      "eval_rougeL": 0.5193148273848067,
+      "eval_rougeLsum": 0.6366010767207634,
+      "eval_runtime": 15.0381,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.26080408692359924,
+      "learning_rate": 0.00018065912780926882,
+      "loss": 1.6591,
+      "step": 3400
+    },
+    {
+      "epoch": 0.39,
+      "eval_bertscore": 0.7327477335929871,
+      "eval_loss": 1.7442594766616821,
+      "eval_rouge1": 0.6424613378037144,
+      "eval_rouge2": 0.34731770322974903,
+      "eval_rougeL": 0.5160705879794565,
+      "eval_rougeLsum": 0.6327006420281607,
+      "eval_runtime": 15.5373,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.23274216055870056,
+      "learning_rate": 0.0001795197538952346,
+      "loss": 1.6613,
+      "step": 3600
+    },
+    {
+      "epoch": 0.41,
+      "eval_bertscore": 0.736956000328064,
+      "eval_loss": 1.7429373264312744,
+      "eval_rouge1": 0.6514574160666677,
+      "eval_rouge2": 0.3556199242231646,
+      "eval_rougeL": 0.5249726237675663,
+      "eval_rougeLsum": 0.6406261097623661,
+      "eval_runtime": 14.9415,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.23616766929626465,
+      "learning_rate": 0.00017838037998120035,
+      "loss": 1.6479,
+      "step": 3800
+    },
+    {
+      "epoch": 0.43,
+      "eval_bertscore": 0.7349627614021301,
+      "eval_loss": 1.7420669794082642,
+      "eval_rouge1": 0.655851684949526,
+      "eval_rouge2": 0.35254590691865084,
+      "eval_rougeL": 0.5248980956621441,
+      "eval_rougeLsum": 0.6449637270581419,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23260319232940674,
+      "learning_rate": 0.0001772410060671661,
+      "loss": 1.6569,
+      "step": 4000
+    },
+    {
+      "epoch": 0.46,
+      "eval_bertscore": 0.7332885265350342,
+      "eval_loss": 1.7401313781738281,
+      "eval_rouge1": 0.6669483634140105,
+      "eval_rouge2": 0.35873988835161297,
+      "eval_rougeL": 0.5343868725007427,
+      "eval_rougeLsum": 0.6555353134690931,
+      "eval_runtime": 15.0822,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2366473525762558,
+      "learning_rate": 0.00017610163215313186,
+      "loss": 1.6599,
+      "step": 4200
+    },
+    {
+      "epoch": 0.48,
+      "eval_bertscore": 0.7335314750671387,
+      "eval_loss": 1.7385823726654053,
+      "eval_rouge1": 0.6559297063578133,
+      "eval_rouge2": 0.35483499789990636,
+      "eval_rougeL": 0.5297939800986089,
+      "eval_rougeLsum": 0.6454544372491222,
+      "eval_runtime": 15.1029,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.20628753304481506,
+      "learning_rate": 0.0001749622582390976,
+      "loss": 1.6454,
+      "step": 4400
+    },
+    {
+      "epoch": 0.5,
+      "eval_bertscore": 0.7342795133590698,
+      "eval_loss": 1.7422058582305908,
+      "eval_rouge1": 0.660746519614568,
+      "eval_rouge2": 0.3633965895561597,
+      "eval_rougeL": 0.5369036980876734,
+      "eval_rougeLsum": 0.650338328328998,
+      "eval_runtime": 15.4434,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 4400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2239149957895279,
+      "learning_rate": 0.0001738228843250634,
+      "loss": 1.6594,
+      "step": 4600
+    },
+    {
+      "epoch": 0.52,
+      "eval_bertscore": 0.7313543558120728,
+      "eval_loss": 1.740854263305664,
+      "eval_rouge1": 0.6591645132619427,
+      "eval_rouge2": 0.35766117432431743,
+      "eval_rougeL": 0.532710255034635,
+      "eval_rougeLsum": 0.6479428185884644,
+      "eval_runtime": 14.9436,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.335,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24808338284492493,
+      "learning_rate": 0.00017268351041102914,
+      "loss": 1.6604,
+      "step": 4800
+    },
+    {
+      "epoch": 0.55,
+      "eval_bertscore": 0.7333321571350098,
+      "eval_loss": 1.7385585308074951,
+      "eval_rouge1": 0.6532115808232871,
+      "eval_rouge2": 0.35333788022501567,
+      "eval_rougeL": 0.5284071547874328,
+      "eval_rougeLsum": 0.6410472452797623,
+      "eval_runtime": 15.0277,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2555364966392517,
+      "learning_rate": 0.0001715441364969949,
+      "loss": 1.6493,
+      "step": 5000
+    },
+    {
+      "epoch": 0.57,
+      "eval_bertscore": 0.7318152189254761,
+      "eval_loss": 1.7357494831085205,
+      "eval_rouge1": 0.6476755890502586,
+      "eval_rouge2": 0.35312778275949164,
+      "eval_rougeL": 0.5227601228049905,
+      "eval_rougeLsum": 0.6371331138372852,
+      "eval_runtime": 14.8807,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 5000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.21518155932426453,
+      "learning_rate": 0.00017040476258296068,
+      "loss": 1.644,
+      "step": 5200
+    },
+    {
+      "epoch": 0.59,
+      "eval_bertscore": 0.734805703163147,
+      "eval_loss": 1.74032723903656,
+      "eval_rouge1": 0.6476813733451636,
+      "eval_rouge2": 0.3509259728617576,
+      "eval_rougeL": 0.5221334872800274,
+      "eval_rougeLsum": 0.636892384667733,
+      "eval_runtime": 15.1271,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.26086658239364624,
+      "learning_rate": 0.00016926538866892643,
+      "loss": 1.6449,
+      "step": 5400
+    },
+    {
+      "epoch": 0.62,
+      "eval_bertscore": 0.7339995503425598,
+      "eval_loss": 1.7338205575942993,
+      "eval_rouge1": 0.6416889902864902,
+      "eval_rouge2": 0.3479045880347737,
+      "eval_rougeL": 0.5160577838468976,
+      "eval_rougeLsum": 0.6317983411796093,
+      "eval_runtime": 14.9992,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 5400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25449469685554504,
+      "learning_rate": 0.0001681260147548922,
+      "loss": 1.6299,
+      "step": 5600
+    },
+    {
+      "epoch": 0.64,
+      "eval_bertscore": 0.7306328415870667,
+      "eval_loss": 1.7369228601455688,
+      "eval_rouge1": 0.6390760905985684,
+      "eval_rouge2": 0.3409328272828699,
+      "eval_rougeL": 0.5111832543685331,
+      "eval_rougeLsum": 0.6285753423407665,
+      "eval_runtime": 15.483,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 5600
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24706102907657623,
+      "learning_rate": 0.00016698664084085796,
+      "loss": 1.6374,
+      "step": 5800
+    },
+    {
+      "epoch": 0.66,
+      "eval_bertscore": 0.732075572013855,
+      "eval_loss": 1.7343876361846924,
+      "eval_rouge1": 0.6378821977913272,
+      "eval_rouge2": 0.34619427775171585,
+      "eval_rougeL": 0.5120186953041237,
+      "eval_rougeLsum": 0.6284056323839109,
+      "eval_runtime": 15.7341,
+      "eval_samples_per_second": 0.636,
+      "eval_steps_per_second": 0.318,
+      "step": 5800
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24373260140419006,
+      "learning_rate": 0.00016584726692682372,
+      "loss": 1.6427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.68,
+      "eval_bertscore": 0.7350374460220337,
+      "eval_loss": 1.729591965675354,
+      "eval_rouge1": 0.6516545226356616,
+      "eval_rouge2": 0.35485762033878543,
+      "eval_rougeL": 0.5249054193354852,
+      "eval_rougeLsum": 0.6411016821651583,
+      "eval_runtime": 15.5199,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 6000
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24616578221321106,
+      "learning_rate": 0.0001647078930127895,
+      "loss": 1.6296,
+      "step": 6200
+    },
+    {
+      "epoch": 0.71,
+      "eval_bertscore": 0.7335461378097534,
+      "eval_loss": 1.7302274703979492,
+      "eval_rouge1": 0.6531411706717427,
+      "eval_rouge2": 0.35003053174601517,
+      "eval_rougeL": 0.5212483686089053,
+      "eval_rougeLsum": 0.6438454124825417,
+      "eval_runtime": 15.3058,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 6200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24500492215156555,
+      "learning_rate": 0.00016356851909875522,
+      "loss": 1.6251,
+      "step": 6400
+    },
+    {
+      "epoch": 0.73,
+      "eval_bertscore": 0.7347471714019775,
+      "eval_loss": 1.7292228937149048,
+      "eval_rouge1": 0.6565322485502556,
+      "eval_rouge2": 0.35887540291607073,
+      "eval_rougeL": 0.5284326132878907,
+      "eval_rougeLsum": 0.6469750895866724,
+      "eval_runtime": 14.9708,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 6400
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.26575228571891785,
+      "learning_rate": 0.000162429145184721,
+      "loss": 1.6389,
+      "step": 6600
+    },
+    {
+      "epoch": 0.75,
+      "eval_bertscore": 0.7319897413253784,
+      "eval_loss": 1.7287580966949463,
+      "eval_rouge1": 0.6458608801881298,
+      "eval_rouge2": 0.3503480901452204,
+      "eval_rougeL": 0.519626708150005,
+      "eval_rougeLsum": 0.6362405734928169,
+      "eval_runtime": 15.3509,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 6600
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.27144965529441833,
+      "learning_rate": 0.00016128977127068676,
+      "loss": 1.6476,
+      "step": 6800
+    },
+    {
+      "epoch": 0.77,
+      "eval_bertscore": 0.7392772436141968,
+      "eval_loss": 1.7257907390594482,
+      "eval_rouge1": 0.6543238897579965,
+      "eval_rouge2": 0.3606726049451984,
+      "eval_rougeL": 0.5317585887753791,
+      "eval_rougeLsum": 0.6452028420624081,
+      "eval_runtime": 15.0268,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 6800
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2579711079597473,
+      "learning_rate": 0.00016015039735665254,
+      "loss": 1.6316,
+      "step": 7000
+    },
+    {
+      "epoch": 0.8,
+      "eval_bertscore": 0.7375612854957581,
+      "eval_loss": 1.7296981811523438,
+      "eval_rouge1": 0.658906725408624,
+      "eval_rouge2": 0.35825094165644866,
+      "eval_rougeL": 0.5323299193377959,
+      "eval_rougeLsum": 0.6500364347290426,
+      "eval_runtime": 14.9244,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 7000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2589207589626312,
+      "learning_rate": 0.0001590110234426183,
+      "loss": 1.6432,
+      "step": 7200
+    },
+    {
+      "epoch": 0.82,
+      "eval_bertscore": 0.735145092010498,
+      "eval_loss": 1.72720205783844,
+      "eval_rouge1": 0.6678646250245518,
+      "eval_rouge2": 0.36332843150846983,
+      "eval_rougeL": 0.537576430733886,
+      "eval_rougeLsum": 0.6579789388660506,
+      "eval_runtime": 14.9067,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.26652559638023376,
+      "learning_rate": 0.00015787164952858404,
+      "loss": 1.6488,
+      "step": 7400
+    },
+    {
+      "epoch": 0.84,
+      "eval_bertscore": 0.7320755124092102,
+      "eval_loss": 1.7282931804656982,
+      "eval_rouge1": 0.6325633297780734,
+      "eval_rouge2": 0.34505856555703185,
+      "eval_rougeL": 0.5100006743383693,
+      "eval_rougeLsum": 0.6230385336341938,
+      "eval_runtime": 14.9075,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 7400
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.27353721857070923,
+      "learning_rate": 0.00015673227561454982,
+      "loss": 1.6486,
+      "step": 7600
+    },
+    {
+      "epoch": 0.87,
+      "eval_bertscore": 0.7367390394210815,
+      "eval_loss": 1.7290785312652588,
+      "eval_rouge1": 0.639487116874423,
+      "eval_rouge2": 0.3466574229736927,
+      "eval_rougeL": 0.515038120249177,
+      "eval_rougeLsum": 0.6301157215372983,
+      "eval_runtime": 15.0876,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.331,
+      "step": 7600
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24777938425540924,
+      "learning_rate": 0.00015559290170051558,
+      "loss": 1.6271,
+      "step": 7800
+    },
+    {
+      "epoch": 0.89,
+      "eval_bertscore": 0.735866904258728,
+      "eval_loss": 1.7264015674591064,
+      "eval_rouge1": 0.64939597901302,
+      "eval_rouge2": 0.3554282813944538,
+      "eval_rougeL": 0.5247953329477759,
+      "eval_rougeLsum": 0.6405524812915908,
+      "eval_runtime": 14.8892,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 7800
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2703794538974762,
+      "learning_rate": 0.00015445352778648136,
+      "loss": 1.6415,
+      "step": 8000
+    },
+    {
+      "epoch": 0.91,
+      "eval_bertscore": 0.7325771450996399,
+      "eval_loss": 1.7271970510482788,
+      "eval_rouge1": 0.6659432894288253,
+      "eval_rouge2": 0.35962933912652617,
+      "eval_rougeL": 0.5385420432813512,
+      "eval_rougeLsum": 0.6557027031484046,
+      "eval_runtime": 14.8319,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 8000
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.28753793239593506,
+      "learning_rate": 0.0001533141538724471,
+      "loss": 1.6239,
+      "step": 8200
+    },
+    {
+      "epoch": 0.93,
+      "eval_bertscore": 0.7350013852119446,
+      "eval_loss": 1.7266199588775635,
+      "eval_rouge1": 0.6549282561414593,
+      "eval_rouge2": 0.35694530595734475,
+      "eval_rougeL": 0.5301601006964574,
+      "eval_rougeLsum": 0.6441779306137909,
+      "eval_runtime": 14.9005,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 8200
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.23870150744915009,
+      "learning_rate": 0.00015217477995841286,
+      "loss": 1.6293,
+      "step": 8400
+    },
+    {
+      "epoch": 0.96,
+      "eval_bertscore": 0.7271261811256409,
+      "eval_loss": 1.7256368398666382,
+      "eval_rouge1": 0.6515513829901936,
+      "eval_rouge2": 0.35217616104918836,
+      "eval_rougeL": 0.5236553509227138,
+      "eval_rougeLsum": 0.6411473505324752,
+      "eval_runtime": 14.9938,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 8400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.28276997804641724,
+      "learning_rate": 0.00015103540604437861,
+      "loss": 1.6242,
+      "step": 8600
+    },
+    {
+      "epoch": 0.98,
+      "eval_bertscore": 0.7347334027290344,
+      "eval_loss": 1.717627763748169,
+      "eval_rouge1": 0.6350112495847634,
+      "eval_rouge2": 0.3477570751550898,
+      "eval_rougeL": 0.5146616989899861,
+      "eval_rougeLsum": 0.6246669376525157,
+      "eval_runtime": 15.7032,
+      "eval_samples_per_second": 0.637,
+      "eval_steps_per_second": 0.318,
+      "step": 8600
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24915842711925507,
+      "learning_rate": 0.00014989603213034437,
+      "loss": 1.6245,
+      "step": 8800
+    },
+    {
+      "epoch": 1.0,
+      "eval_bertscore": 0.7313701510429382,
+      "eval_loss": 1.7292964458465576,
+      "eval_rouge1": 0.6479528105367669,
+      "eval_rouge2": 0.35020983244262877,
+      "eval_rougeL": 0.5200907337780047,
+      "eval_rougeLsum": 0.6372896614836894,
+      "eval_runtime": 15.0043,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 8800
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.24036027491092682,
+      "learning_rate": 0.00014875665821631015,
+      "loss": 1.5364,
+      "step": 9000
+    },
+    {
+      "epoch": 1.03,
+      "eval_bertscore": 0.7327737808227539,
+      "eval_loss": 1.7339435815811157,
+      "eval_rouge1": 0.6568524178922349,
+      "eval_rouge2": 0.35560270713543163,
+      "eval_rougeL": 0.5310443670833082,
+      "eval_rougeLsum": 0.6480993679097387,
+      "eval_runtime": 14.9303,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 9000
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2729027271270752,
+      "learning_rate": 0.0001476172843022759,
+      "loss": 1.5182,
+      "step": 9200
+    },
+    {
+      "epoch": 1.05,
+      "eval_bertscore": 0.7334672212600708,
+      "eval_loss": 1.739061713218689,
+      "eval_rouge1": 0.6552962187824329,
+      "eval_rouge2": 0.35210314124279196,
+      "eval_rougeL": 0.5272039052368354,
+      "eval_rougeLsum": 0.6437473533492806,
+      "eval_runtime": 15.5418,
+      "eval_samples_per_second": 0.643,
+      "eval_steps_per_second": 0.322,
+      "step": 9200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2909716069698334,
+      "learning_rate": 0.00014647791038824168,
+      "loss": 1.5276,
+      "step": 9400
+    },
+    {
+      "epoch": 1.07,
+      "eval_bertscore": 0.7288902997970581,
+      "eval_loss": 1.736944556236267,
+      "eval_rouge1": 0.6533271685598301,
+      "eval_rouge2": 0.35279315321532184,
+      "eval_rougeL": 0.5262688234671329,
+      "eval_rougeLsum": 0.6424084937151033,
+      "eval_runtime": 15.1115,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 9400
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.3035859763622284,
+      "learning_rate": 0.00014533853647420743,
+      "loss": 1.5445,
+      "step": 9600
+    },
+    {
+      "epoch": 1.09,
+      "eval_bertscore": 0.7308284044265747,
+      "eval_loss": 1.737762689590454,
+      "eval_rouge1": 0.6619725777891359,
+      "eval_rouge2": 0.3611963714506864,
+      "eval_rougeL": 0.5363802967084452,
+      "eval_rougeLsum": 0.6516690557971352,
+      "eval_runtime": 14.9932,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 9600
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.26574915647506714,
+      "learning_rate": 0.0001441991625601732,
+      "loss": 1.5342,
+      "step": 9800
+    },
+    {
+      "epoch": 1.12,
+      "eval_bertscore": 0.7328712344169617,
+      "eval_loss": 1.7393991947174072,
+      "eval_rouge1": 0.6856504003396884,
+      "eval_rouge2": 0.3761098841062477,
+      "eval_rougeL": 0.555477293163325,
+      "eval_rougeLsum": 0.6757574283262289,
+      "eval_runtime": 14.8121,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 9800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.315468430519104,
+      "learning_rate": 0.00014305978864613897,
+      "loss": 1.543,
+      "step": 10000
+    },
+    {
+      "epoch": 1.14,
+      "eval_bertscore": 0.7349387407302856,
+      "eval_loss": 1.7352710962295532,
+      "eval_rouge1": 0.6749953128982036,
+      "eval_rouge2": 0.3720385250530084,
+      "eval_rougeL": 0.5472261566474382,
+      "eval_rougeLsum": 0.6657643539219252,
+      "eval_runtime": 14.909,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 10000
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29815027117729187,
+      "learning_rate": 0.00014192041473210472,
+      "loss": 1.5547,
+      "step": 10200
+    },
+    {
+      "epoch": 1.16,
+      "eval_bertscore": 0.7359883189201355,
+      "eval_loss": 1.7269136905670166,
+      "eval_rouge1": 0.6561141614088863,
+      "eval_rouge2": 0.3606175666303814,
+      "eval_rougeL": 0.5302270771032793,
+      "eval_rougeLsum": 0.6446912079521883,
+      "eval_runtime": 14.9527,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 10200
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3595702350139618,
+      "learning_rate": 0.00014078104081807047,
+      "loss": 1.5567,
+      "step": 10400
+    },
+    {
+      "epoch": 1.18,
+      "eval_bertscore": 0.7328116297721863,
+      "eval_loss": 1.7341644763946533,
+      "eval_rouge1": 0.6420332714823549,
+      "eval_rouge2": 0.35094864549032,
+      "eval_rougeL": 0.5179556761398367,
+      "eval_rougeLsum": 0.631987397226809,
+      "eval_runtime": 15.1438,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 10400
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2718666195869446,
+      "learning_rate": 0.00013964166690403623,
+      "loss": 1.5408,
+      "step": 10600
+    },
+    {
+      "epoch": 1.21,
+      "eval_bertscore": 0.7337731122970581,
+      "eval_loss": 1.7330901622772217,
+      "eval_rouge1": 0.661681342484687,
+      "eval_rouge2": 0.3626833509973693,
+      "eval_rougeL": 0.5329424447373774,
+      "eval_rougeLsum": 0.6519750177144633,
+      "eval_runtime": 14.8142,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.338,
+      "step": 10600
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.29183274507522583,
+      "learning_rate": 0.00013850229299000198,
+      "loss": 1.5422,
+      "step": 10800
+    },
+    {
+      "epoch": 1.23,
+      "eval_bertscore": 0.7331051230430603,
+      "eval_loss": 1.7297636270523071,
+      "eval_rouge1": 0.6655497978238063,
+      "eval_rouge2": 0.3614235788441926,
+      "eval_rougeL": 0.5327210061667442,
+      "eval_rougeLsum": 0.6548836840483913,
+      "eval_runtime": 15.1359,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 10800
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.30979740619659424,
+      "learning_rate": 0.00013736291907596776,
+      "loss": 1.5372,
+      "step": 11000
+    },
+    {
+      "epoch": 1.25,
+      "eval_bertscore": 0.7312799692153931,
+      "eval_loss": 1.732444167137146,
+      "eval_rouge1": 0.6568292865033993,
+      "eval_rouge2": 0.35876682221562006,
+      "eval_rougeL": 0.5300878844981931,
+      "eval_rougeLsum": 0.6461751645858989,
+      "eval_runtime": 14.819,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 11000
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.31343138217926025,
+      "learning_rate": 0.0001362235451619335,
+      "loss": 1.5301,
+      "step": 11200
+    },
+    {
+      "epoch": 1.28,
+      "eval_bertscore": 0.7317885160446167,
+      "eval_loss": 1.7358499765396118,
+      "eval_rouge1": 0.6548673097943329,
+      "eval_rouge2": 0.3609116081432997,
+      "eval_rougeL": 0.5279887650752133,
+      "eval_rougeLsum": 0.6466232329097188,
+      "eval_runtime": 14.8259,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36181533336639404,
+      "learning_rate": 0.0001350841712478993,
+      "loss": 1.5421,
+      "step": 11400
+    },
+    {
+      "epoch": 1.3,
+      "eval_bertscore": 0.7316756248474121,
+      "eval_loss": 1.7282969951629639,
+      "eval_rouge1": 0.6551882964480251,
+      "eval_rouge2": 0.3580708921400697,
+      "eval_rougeL": 0.5255367305995147,
+      "eval_rougeLsum": 0.6449192953008009,
+      "eval_runtime": 14.8816,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 11400
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.30600836873054504,
+      "learning_rate": 0.00013394479733386505,
+      "loss": 1.5538,
+      "step": 11600
+    },
+    {
+      "epoch": 1.32,
+      "eval_bertscore": 0.7311854362487793,
+      "eval_loss": 1.7313562631607056,
+      "eval_rouge1": 0.6592751199424156,
+      "eval_rouge2": 0.35802855072854206,
+      "eval_rougeL": 0.5297288176377084,
+      "eval_rougeLsum": 0.6489455314962717,
+      "eval_runtime": 15.0693,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 11600
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29904893040657043,
+      "learning_rate": 0.0001328054234198308,
+      "loss": 1.5328,
+      "step": 11800
+    },
+    {
+      "epoch": 1.34,
+      "eval_bertscore": 0.7312635183334351,
+      "eval_loss": 1.7318429946899414,
+      "eval_rouge1": 0.6577169369077195,
+      "eval_rouge2": 0.3582474830918887,
+      "eval_rougeL": 0.5314990647771975,
+      "eval_rougeLsum": 0.6454785220479866,
+      "eval_runtime": 15.0235,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 11800
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3025416433811188,
+      "learning_rate": 0.00013166604950579658,
+      "loss": 1.5349,
+      "step": 12000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bertscore": 0.7325812578201294,
+      "eval_loss": 1.7309118509292603,
+      "eval_rouge1": 0.6629133074951261,
+      "eval_rouge2": 0.3678158453940578,
+      "eval_rougeL": 0.5380936907276155,
+      "eval_rougeLsum": 0.654883061928214,
+      "eval_runtime": 14.7666,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 12000
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.34982389211654663,
+      "learning_rate": 0.00013052667559176233,
+      "loss": 1.5513,
+      "step": 12200
+    },
+    {
+      "epoch": 1.39,
+      "eval_bertscore": 0.7340582609176636,
+      "eval_loss": 1.7363474369049072,
+      "eval_rouge1": 0.6555817937418287,
+      "eval_rouge2": 0.35630500078358396,
+      "eval_rougeL": 0.5272412353478366,
+      "eval_rougeLsum": 0.6445837479327643,
+      "eval_runtime": 14.9664,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 12200
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.35809043049812317,
+      "learning_rate": 0.0001293873016777281,
+      "loss": 1.5444,
+      "step": 12400
+    },
+    {
+      "epoch": 1.41,
+      "eval_bertscore": 0.7334069013595581,
+      "eval_loss": 1.7328729629516602,
+      "eval_rouge1": 0.6617971237669364,
+      "eval_rouge2": 0.35951260376512423,
+      "eval_rougeL": 0.5345512305507059,
+      "eval_rougeLsum": 0.648363531132752,
+      "eval_runtime": 15.2146,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 12400
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2954196631908417,
+      "learning_rate": 0.00012824792776369387,
+      "loss": 1.5406,
+      "step": 12600
+    },
+    {
+      "epoch": 1.44,
+      "eval_bertscore": 0.7321678400039673,
+      "eval_loss": 1.7335160970687866,
+      "eval_rouge1": 0.6573625593086756,
+      "eval_rouge2": 0.36210525247389347,
+      "eval_rougeL": 0.5379361120230158,
+      "eval_rougeLsum": 0.6459787883452857,
+      "eval_runtime": 14.8942,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 12600
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.32190588116645813,
+      "learning_rate": 0.00012710855384965962,
+      "loss": 1.5491,
+      "step": 12800
+    },
+    {
+      "epoch": 1.46,
+      "eval_bertscore": 0.7346011400222778,
+      "eval_loss": 1.7364966869354248,
+      "eval_rouge1": 0.6481210247390559,
+      "eval_rouge2": 0.3521173896017687,
+      "eval_rougeL": 0.5240500581372636,
+      "eval_rougeLsum": 0.63706442433335,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 12800
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.33323267102241516,
+      "learning_rate": 0.00012596917993562537,
+      "loss": 1.5596,
+      "step": 13000
+    },
+    {
+      "epoch": 1.48,
+      "eval_bertscore": 0.7331587076187134,
+      "eval_loss": 1.7332260608673096,
+      "eval_rouge1": 0.6561257401878793,
+      "eval_rouge2": 0.3548063723792664,
+      "eval_rougeL": 0.527807776001489,
+      "eval_rougeLsum": 0.6451911907984706,
+      "eval_runtime": 15.4703,
+      "eval_samples_per_second": 0.646,
+      "eval_steps_per_second": 0.323,
+      "step": 13000
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3564057946205139,
+      "learning_rate": 0.00012482980602159113,
+      "loss": 1.5261,
+      "step": 13200
+    },
+    {
+      "epoch": 1.5,
+      "eval_bertscore": 0.7306063771247864,
+      "eval_loss": 1.7368810176849365,
+      "eval_rouge1": 0.637722723890071,
+      "eval_rouge2": 0.3455358728458236,
+      "eval_rougeL": 0.5136372690435154,
+      "eval_rougeLsum": 0.6273570573595115,
+      "eval_runtime": 15.3533,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 13200
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29219934344291687,
+      "learning_rate": 0.0001236904321075569,
+      "loss": 1.519,
+      "step": 13400
+    },
+    {
+      "epoch": 1.53,
+      "eval_bertscore": 0.7338696122169495,
+      "eval_loss": 1.734724998474121,
+      "eval_rouge1": 0.6442107446420164,
+      "eval_rouge2": 0.3494748457109431,
+      "eval_rougeL": 0.5207483892007314,
+      "eval_rougeLsum": 0.632886404907802,
+      "eval_runtime": 15.3077,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 13400
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34681758284568787,
+      "learning_rate": 0.00012255105819352266,
+      "loss": 1.5419,
+      "step": 13600
+    },
+    {
+      "epoch": 1.55,
+      "eval_bertscore": 0.7350045442581177,
+      "eval_loss": 1.7329858541488647,
+      "eval_rouge1": 0.6606839869796519,
+      "eval_rouge2": 0.362188561160822,
+      "eval_rougeL": 0.5342033818317451,
+      "eval_rougeLsum": 0.6493340000068861,
+      "eval_runtime": 15.44,
+      "eval_samples_per_second": 0.648,
+      "eval_steps_per_second": 0.324,
+      "step": 13600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3043666481971741,
+      "learning_rate": 0.00012141168427948844,
+      "loss": 1.5402,
+      "step": 13800
+    },
+    {
+      "epoch": 1.57,
+      "eval_bertscore": 0.7363221645355225,
+      "eval_loss": 1.7308530807495117,
+      "eval_rouge1": 0.6638252384356028,
+      "eval_rouge2": 0.3643237697892826,
+      "eval_rougeL": 0.5403775887381331,
+      "eval_rougeLsum": 0.6537260000827279,
+      "eval_runtime": 14.7668,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.339,
+      "step": 13800
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4073585867881775,
+      "learning_rate": 0.00012027231036545419,
+      "loss": 1.5256,
+      "step": 14000
+    },
+    {
+      "epoch": 1.59,
+      "eval_bertscore": 0.7310279607772827,
+      "eval_loss": 1.7326784133911133,
+      "eval_rouge1": 0.6609594314120198,
+      "eval_rouge2": 0.3601530714440473,
+      "eval_rougeL": 0.5344452687135626,
+      "eval_rougeLsum": 0.6480936554342305,
+      "eval_runtime": 14.8565,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 14000
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3211813271045685,
+      "learning_rate": 0.00011913293645141995,
+      "loss": 1.5366,
+      "step": 14200
+    },
+    {
+      "epoch": 1.62,
+      "eval_bertscore": 0.7356667518615723,
+      "eval_loss": 1.7280094623565674,
+      "eval_rouge1": 0.6519353227375031,
+      "eval_rouge2": 0.3587025716186173,
+      "eval_rougeL": 0.5306356200586075,
+      "eval_rougeLsum": 0.6408870347994059,
+      "eval_runtime": 14.9264,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 14200
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.32776832580566406,
+      "learning_rate": 0.00011799356253738571,
+      "loss": 1.5504,
+      "step": 14400
+    },
+    {
+      "epoch": 1.64,
+      "eval_bertscore": 0.7331353425979614,
+      "eval_loss": 1.7308950424194336,
+      "eval_rouge1": 0.6627702292814652,
+      "eval_rouge2": 0.36117793957379707,
+      "eval_rougeL": 0.5369305446079228,
+      "eval_rougeLsum": 0.6516924083980089,
+      "eval_runtime": 16.0138,
+      "eval_samples_per_second": 0.624,
+      "eval_steps_per_second": 0.312,
+      "step": 14400
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3209726810455322,
+      "learning_rate": 0.00011685418862335147,
+      "loss": 1.5473,
+      "step": 14600
+    },
+    {
+      "epoch": 1.66,
+      "eval_bertscore": 0.732498824596405,
+      "eval_loss": 1.7328402996063232,
+      "eval_rouge1": 0.6482679740803596,
+      "eval_rouge2": 0.3538726087405498,
+      "eval_rougeL": 0.5267677183598017,
+      "eval_rougeLsum": 0.6366529460029322,
+      "eval_runtime": 15.0195,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 14600
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.3174591064453125,
+      "learning_rate": 0.00011571481470931725,
+      "loss": 1.5568,
+      "step": 14800
+    },
+    {
+      "epoch": 1.69,
+      "eval_bertscore": 0.7335298657417297,
+      "eval_loss": 1.7310253381729126,
+      "eval_rouge1": 0.6560468577439627,
+      "eval_rouge2": 0.36039371229175,
+      "eval_rougeL": 0.5318708569729291,
+      "eval_rougeLsum": 0.6444857558837042,
+      "eval_runtime": 14.9774,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 14800
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2936408817768097,
+      "learning_rate": 0.000114575440795283,
+      "loss": 1.5345,
+      "step": 15000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bertscore": 0.7322725057601929,
+      "eval_loss": 1.7270629405975342,
+      "eval_rouge1": 0.6387060930656672,
+      "eval_rouge2": 0.3480508127989137,
+      "eval_rougeL": 0.5148670834213287,
+      "eval_rougeLsum": 0.6273654952601909,
+      "eval_runtime": 15.6687,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 15000
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.32960689067840576,
+      "learning_rate": 0.00011343606688124875,
+      "loss": 1.5362,
+      "step": 15200
+    },
+    {
+      "epoch": 1.73,
+      "eval_bertscore": 0.7337037920951843,
+      "eval_loss": 1.7287395000457764,
+      "eval_rouge1": 0.6476816970229771,
+      "eval_rouge2": 0.3532248216683249,
+      "eval_rougeL": 0.5253136618838716,
+      "eval_rougeLsum": 0.6347493764394183,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15200
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.33265602588653564,
+      "learning_rate": 0.00011229669296721452,
+      "loss": 1.5215,
+      "step": 15400
+    },
+    {
+      "epoch": 1.75,
+      "eval_bertscore": 0.7330806851387024,
+      "eval_loss": 1.7265052795410156,
+      "eval_rouge1": 0.6529393512177359,
+      "eval_rouge2": 0.36182153145062224,
+      "eval_rougeL": 0.5317061134915853,
+      "eval_rougeLsum": 0.6413066299251913,
+      "eval_runtime": 15.0256,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 15400
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3436201512813568,
+      "learning_rate": 0.00011115731905318027,
+      "loss": 1.539,
+      "step": 15600
+    },
+    {
+      "epoch": 1.78,
+      "eval_bertscore": 0.7335551977157593,
+      "eval_loss": 1.7254730463027954,
+      "eval_rouge1": 0.6388518781767971,
+      "eval_rouge2": 0.3501853846588857,
+      "eval_rougeL": 0.5196828245794569,
+      "eval_rougeLsum": 0.629333993884722,
+      "eval_runtime": 15.2595,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.328,
+      "step": 15600
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3428190350532532,
+      "learning_rate": 0.00011001794513914605,
+      "loss": 1.5273,
+      "step": 15800
+    },
+    {
+      "epoch": 1.8,
+      "eval_bertscore": 0.7331770658493042,
+      "eval_loss": 1.7286018133163452,
+      "eval_rouge1": 0.6581941047310954,
+      "eval_rouge2": 0.36277983926897583,
+      "eval_rougeL": 0.5336464680120501,
+      "eval_rougeLsum": 0.6489239720278894,
+      "eval_runtime": 14.8252,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 15800
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.363164484500885,
+      "learning_rate": 0.0001088785712251118,
+      "loss": 1.5445,
+      "step": 16000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bertscore": 0.7377282977104187,
+      "eval_loss": 1.7363064289093018,
+      "eval_rouge1": 0.6547011011872876,
+      "eval_rouge2": 0.3553220826957326,
+      "eval_rougeL": 0.5256073814411315,
+      "eval_rougeLsum": 0.6420095316923398,
+      "eval_runtime": 14.8599,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 16000
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3098333775997162,
+      "learning_rate": 0.00010773919731107757,
+      "loss": 1.5319,
+      "step": 16200
+    },
+    {
+      "epoch": 1.85,
+      "eval_bertscore": 0.7324053645133972,
+      "eval_loss": 1.7284066677093506,
+      "eval_rouge1": 0.6477379941950916,
+      "eval_rouge2": 0.3535918140554809,
+      "eval_rougeL": 0.5226838544730126,
+      "eval_rougeLsum": 0.6373271915355557,
+      "eval_runtime": 14.9318,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 16200
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.3637208938598633,
+      "learning_rate": 0.00010659982339704332,
+      "loss": 1.5442,
+      "step": 16400
+    },
+    {
+      "epoch": 1.87,
+      "eval_bertscore": 0.7347462773323059,
+      "eval_loss": 1.7252963781356812,
+      "eval_rouge1": 0.6494449840449819,
+      "eval_rouge2": 0.3586550050575282,
+      "eval_rougeL": 0.5275675395159809,
+      "eval_rougeLsum": 0.6396738714026391,
+      "eval_runtime": 15.2218,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 16400
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.35197457671165466,
+      "learning_rate": 0.00010546044948300908,
+      "loss": 1.5131,
+      "step": 16600
+    },
+    {
+      "epoch": 1.89,
+      "eval_bertscore": 0.7329785227775574,
+      "eval_loss": 1.7285687923431396,
+      "eval_rouge1": 0.6582047811143328,
+      "eval_rouge2": 0.3637700686094697,
+      "eval_rougeL": 0.5355021948480279,
+      "eval_rougeLsum": 0.6483245595148677,
+      "eval_runtime": 14.8772,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 16600
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3406757116317749,
+      "learning_rate": 0.00010432107556897486,
+      "loss": 1.5394,
+      "step": 16800
+    },
+    {
+      "epoch": 1.91,
+      "eval_bertscore": 0.7345961332321167,
+      "eval_loss": 1.7324028015136719,
+      "eval_rouge1": 0.6408293615351552,
+      "eval_rouge2": 0.3520120690778129,
+      "eval_rougeL": 0.5145218014745592,
+      "eval_rougeLsum": 0.6297802607384266,
+      "eval_runtime": 15.1044,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 16800
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3417683243751526,
+      "learning_rate": 0.00010318170165494061,
+      "loss": 1.526,
+      "step": 17000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bertscore": 0.735752522945404,
+      "eval_loss": 1.7288110256195068,
+      "eval_rouge1": 0.641158513352794,
+      "eval_rouge2": 0.3544166440855814,
+      "eval_rougeL": 0.5215201980495414,
+      "eval_rougeLsum": 0.630550065494593,
+      "eval_runtime": 15.0797,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 17000
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3256611227989197,
+      "learning_rate": 0.00010204232774090639,
+      "loss": 1.5484,
+      "step": 17200
+    },
+    {
+      "epoch": 1.96,
+      "eval_bertscore": 0.7356327772140503,
+      "eval_loss": 1.7305186986923218,
+      "eval_rouge1": 0.6400269226515611,
+      "eval_rouge2": 0.3502884634173268,
+      "eval_rougeL": 0.517312321281175,
+      "eval_rougeLsum": 0.6284556997614409,
+      "eval_runtime": 15.4097,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 17200
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.4035187363624573,
+      "learning_rate": 0.00010090295382687213,
+      "loss": 1.5261,
+      "step": 17400
+    },
+    {
+      "epoch": 1.98,
+      "eval_bertscore": 0.7339992523193359,
+      "eval_loss": 1.7282793521881104,
+      "eval_rouge1": 0.6335770390416183,
+      "eval_rouge2": 0.34592404578075897,
+      "eval_rougeL": 0.5109045259792113,
+      "eval_rougeLsum": 0.6218413683710426,
+      "eval_runtime": 15.1959,
+      "eval_samples_per_second": 0.658,
+      "eval_steps_per_second": 0.329,
+      "step": 17400
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34987062215805054,
+      "learning_rate": 9.97635799128379e-05,
+      "loss": 1.5199,
+      "step": 17600
+    },
+    {
+      "epoch": 2.0,
+      "eval_bertscore": 0.7326329946517944,
+      "eval_loss": 1.7544715404510498,
+      "eval_rouge1": 0.6451558045750205,
+      "eval_rouge2": 0.35565806935653943,
+      "eval_rougeL": 0.5217034865840529,
+      "eval_rougeLsum": 0.6329869715356753,
+      "eval_runtime": 15.0351,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 17600
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.37184038758277893,
+      "learning_rate": 9.862420599880366e-05,
+      "loss": 1.41,
+      "step": 17800
+    },
+    {
+      "epoch": 2.03,
+      "eval_bertscore": 0.7315141558647156,
+      "eval_loss": 1.7585878372192383,
+      "eval_rouge1": 0.6469319193583706,
+      "eval_rouge2": 0.3514447211469598,
+      "eval_rougeL": 0.524755857688278,
+      "eval_rougeLsum": 0.6350164781858667,
+      "eval_runtime": 14.9583,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 17800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3812776803970337,
+      "learning_rate": 9.748483208476943e-05,
+      "loss": 1.4132,
+      "step": 18000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bertscore": 0.7335561513900757,
+      "eval_loss": 1.764611840248108,
+      "eval_rouge1": 0.6381916780473581,
+      "eval_rouge2": 0.3482510604092539,
+      "eval_rougeL": 0.5162105225823392,
+      "eval_rougeLsum": 0.627150245441782,
+      "eval_runtime": 15.8515,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 18000
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.45525220036506653,
+      "learning_rate": 9.634545817073518e-05,
+      "loss": 1.4,
+      "step": 18200
+    },
+    {
+      "epoch": 2.07,
+      "eval_bertscore": 0.73627769947052,
+      "eval_loss": 1.7585163116455078,
+      "eval_rouge1": 0.6670097658027134,
+      "eval_rouge2": 0.3658295359911405,
+      "eval_rougeL": 0.5429667657900548,
+      "eval_rougeLsum": 0.6543501745791419,
+      "eval_runtime": 15.0301,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 18200
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.37322184443473816,
+      "learning_rate": 9.520608425670095e-05,
+      "loss": 1.4293,
+      "step": 18400
+    },
+    {
+      "epoch": 2.1,
+      "eval_bertscore": 0.730435848236084,
+      "eval_loss": 1.764052391052246,
+      "eval_rouge1": 0.6640215078213034,
+      "eval_rouge2": 0.3625932287322054,
+      "eval_rougeL": 0.5379978391335138,
+      "eval_rougeLsum": 0.6542054656293199,
+      "eval_runtime": 15.0762,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 18400
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4260891079902649,
+      "learning_rate": 9.40667103426667e-05,
+      "loss": 1.4077,
+      "step": 18600
+    },
+    {
+      "epoch": 2.12,
+      "eval_bertscore": 0.7309869527816772,
+      "eval_loss": 1.762108564376831,
+      "eval_rouge1": 0.6571171081737958,
+      "eval_rouge2": 0.35780421333141865,
+      "eval_rougeL": 0.5320129270967632,
+      "eval_rougeLsum": 0.64587787409523,
+      "eval_runtime": 14.9004,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.336,
+      "step": 18600
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.39479926228523254,
+      "learning_rate": 9.292733642863247e-05,
+      "loss": 1.4165,
+      "step": 18800
+    },
+    {
+      "epoch": 2.14,
+      "eval_bertscore": 0.7324444651603699,
+      "eval_loss": 1.7607113122940063,
+      "eval_rouge1": 0.6628398862884018,
+      "eval_rouge2": 0.3627259806721216,
+      "eval_rougeL": 0.5366106483832656,
+      "eval_rougeLsum": 0.6528364858807157,
+      "eval_runtime": 15.5766,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 18800
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.39267703890800476,
+      "learning_rate": 9.178796251459824e-05,
+      "loss": 1.4123,
+      "step": 19000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bertscore": 0.7298994064331055,
+      "eval_loss": 1.7668545246124268,
+      "eval_rouge1": 0.6490850022857569,
+      "eval_rouge2": 0.3532323419511264,
+      "eval_rougeL": 0.5212823000193295,
+      "eval_rougeLsum": 0.636442724466695,
+      "eval_runtime": 14.9094,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 19000
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.38221287727355957,
+      "learning_rate": 9.0648588600564e-05,
+      "loss": 1.401,
+      "step": 19200
+    },
+    {
+      "epoch": 2.19,
+      "eval_bertscore": 0.7316875457763672,
+      "eval_loss": 1.764147400856018,
+      "eval_rouge1": 0.6490326710625849,
+      "eval_rouge2": 0.3510351037900723,
+      "eval_rougeL": 0.5239165028795836,
+      "eval_rougeLsum": 0.6373687316421427,
+      "eval_runtime": 15.1192,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 19200
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.3653150200843811,
+      "learning_rate": 8.950921468652976e-05,
+      "loss": 1.4109,
+      "step": 19400
+    },
+    {
+      "epoch": 2.21,
+      "eval_bertscore": 0.7348155975341797,
+      "eval_loss": 1.7640550136566162,
+      "eval_rouge1": 0.6462152873276823,
+      "eval_rouge2": 0.3483599145461069,
+      "eval_rougeL": 0.5193372430687719,
+      "eval_rougeLsum": 0.6334254357511564,
+      "eval_runtime": 14.9291,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 19400
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.38049009442329407,
+      "learning_rate": 8.836984077249551e-05,
+      "loss": 1.4189,
+      "step": 19600
+    },
+    {
+      "epoch": 2.23,
+      "eval_bertscore": 0.7357938885688782,
+      "eval_loss": 1.7696326971054077,
+      "eval_rouge1": 0.6377276221057538,
+      "eval_rouge2": 0.3455397190390045,
+      "eval_rougeL": 0.5118069428064842,
+      "eval_rougeLsum": 0.6264501633078481,
+      "eval_runtime": 14.8653,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.336,
+      "step": 19600
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.42111098766326904,
+      "learning_rate": 8.723046685846128e-05,
+      "loss": 1.4152,
+      "step": 19800
+    },
+    {
+      "epoch": 2.26,
+      "eval_bertscore": 0.7339056134223938,
+      "eval_loss": 1.7658218145370483,
+      "eval_rouge1": 0.6494820372695989,
+      "eval_rouge2": 0.34691658128805236,
+      "eval_rougeL": 0.5193228965163086,
+      "eval_rougeLsum": 0.6365347065687565,
+      "eval_runtime": 15.3562,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 19800
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.4452258050441742,
+      "learning_rate": 8.609109294442704e-05,
+      "loss": 1.4101,
+      "step": 20000
+    },
+    {
+      "epoch": 2.28,
+      "eval_bertscore": 0.7296434640884399,
+      "eval_loss": 1.7714240550994873,
+      "eval_rouge1": 0.6565600824405751,
+      "eval_rouge2": 0.3533618655201594,
+      "eval_rougeL": 0.5263318202066467,
+      "eval_rougeLsum": 0.6444964824298407,
+      "eval_runtime": 14.8578,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 20000
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4030967652797699,
+      "learning_rate": 8.495171903039281e-05,
+      "loss": 1.4049,
+      "step": 20200
+    },
+    {
+      "epoch": 2.3,
+      "eval_bertscore": 0.7307097315788269,
+      "eval_loss": 1.774444580078125,
+      "eval_rouge1": 0.6517204836155526,
+      "eval_rouge2": 0.3521339653276223,
+      "eval_rougeL": 0.5223211728244184,
+      "eval_rougeLsum": 0.6398710531932736,
+      "eval_runtime": 15.8543,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 20200
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.33409813046455383,
+      "learning_rate": 8.381234511635858e-05,
+      "loss": 1.4243,
+      "step": 20400
+    },
+    {
+      "epoch": 2.32,
+      "eval_bertscore": 0.7312101721763611,
+      "eval_loss": 1.7654094696044922,
+      "eval_rouge1": 0.6607249126293291,
+      "eval_rouge2": 0.3545993249716188,
+      "eval_rougeL": 0.5320161007986739,
+      "eval_rougeLsum": 0.6503315335963733,
+      "eval_runtime": 14.8739,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 20400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.4044789671897888,
+      "learning_rate": 8.267297120232433e-05,
+      "loss": 1.413,
+      "step": 20600
+    },
+    {
+      "epoch": 2.35,
+      "eval_bertscore": 0.7342169880867004,
+      "eval_loss": 1.769879937171936,
+      "eval_rouge1": 0.6442777880355144,
+      "eval_rouge2": 0.35006080708477183,
+      "eval_rougeL": 0.5218799478770955,
+      "eval_rougeLsum": 0.6332700294558067,
+      "eval_runtime": 14.9089,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 20600
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39801183342933655,
+      "learning_rate": 8.153359728829008e-05,
+      "loss": 1.4177,
+      "step": 20800
+    },
+    {
+      "epoch": 2.37,
+      "eval_bertscore": 0.7343758344650269,
+      "eval_loss": 1.7737929821014404,
+      "eval_rouge1": 0.6495678172205896,
+      "eval_rouge2": 0.3505195734345703,
+      "eval_rougeL": 0.5263025592812188,
+      "eval_rougeLsum": 0.6390057749428748,
+      "eval_runtime": 15.2148,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.329,
+      "step": 20800
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.36868759989738464,
+      "learning_rate": 8.039422337425585e-05,
+      "loss": 1.421,
+      "step": 21000
+    },
+    {
+      "epoch": 2.39,
+      "eval_bertscore": 0.7333502173423767,
+      "eval_loss": 1.7708820104599,
+      "eval_rouge1": 0.656319412860679,
+      "eval_rouge2": 0.3557406341135577,
+      "eval_rougeL": 0.5293456110466322,
+      "eval_rougeLsum": 0.6421819358163285,
+      "eval_runtime": 14.9091,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 21000
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.46111443638801575,
+      "learning_rate": 7.925484946022162e-05,
+      "loss": 1.4102,
+      "step": 21200
+    },
+    {
+      "epoch": 2.41,
+      "eval_bertscore": 0.736262321472168,
+      "eval_loss": 1.768972635269165,
+      "eval_rouge1": 0.6582574071278393,
+      "eval_rouge2": 0.3557625250443591,
+      "eval_rougeL": 0.5322500342922363,
+      "eval_rougeLsum": 0.646623827844921,
+      "eval_runtime": 15.4088,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21200
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.41794517636299133,
+      "learning_rate": 7.811547554618738e-05,
+      "loss": 1.4231,
+      "step": 21400
+    },
+    {
+      "epoch": 2.44,
+      "eval_bertscore": 0.7349900603294373,
+      "eval_loss": 1.7673609256744385,
+      "eval_rouge1": 0.6599777278993147,
+      "eval_rouge2": 0.35744569380532043,
+      "eval_rougeL": 0.5359850821835463,
+      "eval_rougeLsum": 0.6469206354455653,
+      "eval_runtime": 14.8837,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 21400
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3874039351940155,
+      "learning_rate": 7.697610163215314e-05,
+      "loss": 1.4158,
+      "step": 21600
+    },
+    {
+      "epoch": 2.46,
+      "eval_bertscore": 0.7362676858901978,
+      "eval_loss": 1.764347791671753,
+      "eval_rouge1": 0.6576168971663054,
+      "eval_rouge2": 0.36010190798950537,
+      "eval_rougeL": 0.5365592740576962,
+      "eval_rougeLsum": 0.6455601225938818,
+      "eval_runtime": 15.4178,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 21600
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.4013253450393677,
+      "learning_rate": 7.583672771811889e-05,
+      "loss": 1.418,
+      "step": 21800
+    },
+    {
+      "epoch": 2.48,
+      "eval_bertscore": 0.7303592562675476,
+      "eval_loss": 1.765144944190979,
+      "eval_rouge1": 0.6610480012685163,
+      "eval_rouge2": 0.36479831105715255,
+      "eval_rougeL": 0.5375415216439376,
+      "eval_rougeLsum": 0.6504955320897916,
+      "eval_runtime": 15.0677,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 21800
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.4189004898071289,
+      "learning_rate": 7.469735380408466e-05,
+      "loss": 1.4199,
+      "step": 22000
+    },
+    {
+      "epoch": 2.51,
+      "eval_bertscore": 0.7319179773330688,
+      "eval_loss": 1.7685811519622803,
+      "eval_rouge1": 0.6589314825298751,
+      "eval_rouge2": 0.36092809773515727,
+      "eval_rougeL": 0.5359034256928837,
+      "eval_rougeLsum": 0.647695652924998,
+      "eval_runtime": 15.0325,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 22000
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.39296436309814453,
+      "learning_rate": 7.355797989005042e-05,
+      "loss": 1.4353,
+      "step": 22200
+    },
+    {
+      "epoch": 2.53,
+      "eval_bertscore": 0.734102725982666,
+      "eval_loss": 1.7720457315444946,
+      "eval_rouge1": 0.652136441919871,
+      "eval_rouge2": 0.35394856883334874,
+      "eval_rougeL": 0.5257845140699575,
+      "eval_rougeLsum": 0.6411232244792167,
+      "eval_runtime": 14.9943,
+      "eval_samples_per_second": 0.667,
+      "eval_steps_per_second": 0.333,
+      "step": 22200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.3997296392917633,
+      "learning_rate": 7.241860597601619e-05,
+      "loss": 1.4224,
+      "step": 22400
+    },
+    {
+      "epoch": 2.55,
+      "eval_bertscore": 0.7339878082275391,
+      "eval_loss": 1.7666336297988892,
+      "eval_rouge1": 0.6537340121878514,
+      "eval_rouge2": 0.3570961026063757,
+      "eval_rougeL": 0.529937130767685,
+      "eval_rougeLsum": 0.6435060914147177,
+      "eval_runtime": 15.0439,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 22400
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.45447298884391785,
+      "learning_rate": 7.127923206198196e-05,
+      "loss": 1.4195,
+      "step": 22600
+    },
+    {
+      "epoch": 2.57,
+      "eval_bertscore": 0.7328049540519714,
+      "eval_loss": 1.767970085144043,
+      "eval_rouge1": 0.6525615870662755,
+      "eval_rouge2": 0.3548658659692201,
+      "eval_rougeL": 0.5278612681579985,
+      "eval_rougeLsum": 0.6426018669254849,
+      "eval_runtime": 14.8509,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 22600
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.37068402767181396,
+      "learning_rate": 7.01398581479477e-05,
+      "loss": 1.4174,
+      "step": 22800
+    },
+    {
+      "epoch": 2.6,
+      "eval_bertscore": 0.7338019609451294,
+      "eval_loss": 1.763349175453186,
+      "eval_rouge1": 0.6536530437395975,
+      "eval_rouge2": 0.3564778360043106,
+      "eval_rougeL": 0.5285382022264695,
+      "eval_rougeLsum": 0.6425723229746058,
+      "eval_runtime": 14.8293,
+      "eval_samples_per_second": 0.674,
+      "eval_steps_per_second": 0.337,
+      "step": 22800
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.35103845596313477,
+      "learning_rate": 6.900048423391346e-05,
+      "loss": 1.4176,
+      "step": 23000
+    },
+    {
+      "epoch": 2.62,
+      "eval_bertscore": 0.7325159311294556,
+      "eval_loss": 1.7628978490829468,
+      "eval_rouge1": 0.6623030759220351,
+      "eval_rouge2": 0.3604109991839185,
+      "eval_rougeL": 0.5322677462077166,
+      "eval_rougeLsum": 0.6516312160764892,
+      "eval_runtime": 14.923,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 23000
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.39424487948417664,
+      "learning_rate": 6.786111031987923e-05,
+      "loss": 1.4141,
+      "step": 23200
+    },
+    {
+      "epoch": 2.64,
+      "eval_bertscore": 0.7374362945556641,
+      "eval_loss": 1.7650716304779053,
+      "eval_rouge1": 0.6574628250156043,
+      "eval_rouge2": 0.36049809448726045,
+      "eval_rougeL": 0.5312753147070929,
+      "eval_rougeLsum": 0.6452805224085838,
+      "eval_runtime": 14.8888,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 23200
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.43388617038726807,
+      "learning_rate": 6.6721736405845e-05,
+      "loss": 1.4162,
+      "step": 23400
+    },
+    {
+      "epoch": 2.67,
+      "eval_bertscore": 0.7336270809173584,
+      "eval_loss": 1.7658464908599854,
+      "eval_rouge1": 0.6573821610105346,
+      "eval_rouge2": 0.3554905858433707,
+      "eval_rougeL": 0.5301171280694805,
+      "eval_rougeLsum": 0.6452831484327366,
+      "eval_runtime": 14.9065,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 23400
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.41646161675453186,
+      "learning_rate": 6.558236249181076e-05,
+      "loss": 1.4199,
+      "step": 23600
+    },
+    {
+      "epoch": 2.69,
+      "eval_bertscore": 0.7336153388023376,
+      "eval_loss": 1.767188310623169,
+      "eval_rouge1": 0.6571676655188523,
+      "eval_rouge2": 0.3572385458490631,
+      "eval_rougeL": 0.5296070447249894,
+      "eval_rougeLsum": 0.6457403843177614,
+      "eval_runtime": 14.7731,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 23600
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.38818874955177307,
+      "learning_rate": 6.444298857777651e-05,
+      "loss": 1.416,
+      "step": 23800
+    },
+    {
+      "epoch": 2.71,
+      "eval_bertscore": 0.7327283620834351,
+      "eval_loss": 1.7616260051727295,
+      "eval_rouge1": 0.656883853658625,
+      "eval_rouge2": 0.3566053754349987,
+      "eval_rougeL": 0.5314518531110131,
+      "eval_rougeLsum": 0.6446071370588691,
+      "eval_runtime": 15.1059,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 23800
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.39458197355270386,
+      "learning_rate": 6.330361466374227e-05,
+      "loss": 1.4243,
+      "step": 24000
+    },
+    {
+      "epoch": 2.73,
+      "eval_bertscore": 0.7317964434623718,
+      "eval_loss": 1.7638084888458252,
+      "eval_rouge1": 0.6571606334248314,
+      "eval_rouge2": 0.3569373186268652,
+      "eval_rougeL": 0.5293554693809414,
+      "eval_rougeLsum": 0.645057709779701,
+      "eval_runtime": 14.8582,
+      "eval_samples_per_second": 0.673,
+      "eval_steps_per_second": 0.337,
+      "step": 24000
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.38834938406944275,
+      "learning_rate": 6.216424074970803e-05,
+      "loss": 1.4131,
+      "step": 24200
+    },
+    {
+      "epoch": 2.76,
+      "eval_bertscore": 0.7318626642227173,
+      "eval_loss": 1.7657394409179688,
+      "eval_rouge1": 0.6537968135786872,
+      "eval_rouge2": 0.3558568019037358,
+      "eval_rougeL": 0.528787979657496,
+      "eval_rougeLsum": 0.643715855457879,
+      "eval_runtime": 14.7776,
+      "eval_samples_per_second": 0.677,
+      "eval_steps_per_second": 0.338,
+      "step": 24200
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.37283340096473694,
+      "learning_rate": 6.10248668356738e-05,
+      "loss": 1.4203,
+      "step": 24400
+    },
+    {
+      "epoch": 2.78,
+      "eval_bertscore": 0.7367373704910278,
+      "eval_loss": 1.7650487422943115,
+      "eval_rouge1": 0.6541794402701134,
+      "eval_rouge2": 0.3544283778817331,
+      "eval_rougeL": 0.5276595078021217,
+      "eval_rougeLsum": 0.641985089875446,
+      "eval_runtime": 15.2752,
+      "eval_samples_per_second": 0.655,
+      "eval_steps_per_second": 0.327,
+      "step": 24400
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4221261441707611,
+      "learning_rate": 5.988549292163956e-05,
+      "loss": 1.4207,
+      "step": 24600
+    },
+    {
+      "epoch": 2.8,
+      "eval_bertscore": 0.7366531491279602,
+      "eval_loss": 1.7622770071029663,
+      "eval_rouge1": 0.6593934034597051,
+      "eval_rouge2": 0.359831500847421,
+      "eval_rougeL": 0.5337810293575901,
+      "eval_rougeLsum": 0.647676431170948,
+      "eval_runtime": 16.3422,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 24600
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.4765646457672119,
+      "learning_rate": 5.874611900760533e-05,
+      "loss": 1.4105,
+      "step": 24800
+    },
+    {
+      "epoch": 2.83,
+      "eval_bertscore": 0.7325325012207031,
+      "eval_loss": 1.7692537307739258,
+      "eval_rouge1": 0.6507634466188938,
+      "eval_rouge2": 0.3518723325059453,
+      "eval_rougeL": 0.5245280386580646,
+      "eval_rougeLsum": 0.6407163362773103,
+      "eval_runtime": 15.8625,
+      "eval_samples_per_second": 0.63,
+      "eval_steps_per_second": 0.315,
+      "step": 24800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.43420910835266113,
+      "learning_rate": 5.760674509357109e-05,
+      "loss": 1.4093,
+      "step": 25000
+    },
+    {
+      "epoch": 2.85,
+      "eval_bertscore": 0.7350980043411255,
+      "eval_loss": 1.765899896621704,
+      "eval_rouge1": 0.654968133779226,
+      "eval_rouge2": 0.35640798019371084,
+      "eval_rougeL": 0.5293198154872452,
+      "eval_rougeLsum": 0.6438911452718419,
+      "eval_runtime": 14.9802,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 25000
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.4037187993526459,
+      "learning_rate": 5.646737117953684e-05,
+      "loss": 1.4168,
+      "step": 25200
+    },
+    {
+      "epoch": 2.87,
+      "eval_bertscore": 0.7334460020065308,
+      "eval_loss": 1.7674219608306885,
+      "eval_rouge1": 0.6449835105373629,
+      "eval_rouge2": 0.3523964707950204,
+      "eval_rougeL": 0.5229607326177911,
+      "eval_rougeLsum": 0.6341183290076015,
+      "eval_runtime": 15.1444,
+      "eval_samples_per_second": 0.66,
+      "eval_steps_per_second": 0.33,
+      "step": 25200
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.46987786889076233,
+      "learning_rate": 5.532799726550261e-05,
+      "loss": 1.4232,
+      "step": 25400
+    },
+    {
+      "epoch": 2.89,
+      "eval_bertscore": 0.7338739037513733,
+      "eval_loss": 1.766469955444336,
+      "eval_rouge1": 0.652936115282645,
+      "eval_rouge2": 0.35620333501115786,
+      "eval_rougeL": 0.5268542487630437,
+      "eval_rougeLsum": 0.6430711603020087,
+      "eval_runtime": 15.1688,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.33,
+      "step": 25400
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.3969995081424713,
+      "learning_rate": 5.418862335146837e-05,
+      "loss": 1.4134,
+      "step": 25600
+    },
+    {
+      "epoch": 2.92,
+      "eval_bertscore": 0.7372814416885376,
+      "eval_loss": 1.7634046077728271,
+      "eval_rouge1": 0.6612907782823452,
+      "eval_rouge2": 0.3600341319079672,
+      "eval_rougeL": 0.5358216104379979,
+      "eval_rougeLsum": 0.6517186363306726,
+      "eval_runtime": 15.3081,
+      "eval_samples_per_second": 0.653,
+      "eval_steps_per_second": 0.327,
+      "step": 25600
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.4024534821510315,
+      "learning_rate": 5.3049249437434134e-05,
+      "loss": 1.4101,
+      "step": 25800
+    },
+    {
+      "epoch": 2.94,
+      "eval_bertscore": 0.7351590991020203,
+      "eval_loss": 1.762956976890564,
+      "eval_rouge1": 0.6564585752882299,
+      "eval_rouge2": 0.35673018577153914,
+      "eval_rougeL": 0.5313007749373967,
+      "eval_rougeLsum": 0.6469154732294412,
+      "eval_runtime": 15.373,
+      "eval_samples_per_second": 0.65,
+      "eval_steps_per_second": 0.325,
+      "step": 25800
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3964459300041199,
+      "learning_rate": 5.19098755233999e-05,
+      "loss": 1.4238,
+      "step": 26000
+    },
+    {
+      "epoch": 2.96,
+      "eval_bertscore": 0.7316532135009766,
+      "eval_loss": 1.7618186473846436,
+      "eval_rouge1": 0.654963067591155,
+      "eval_rouge2": 0.3581644146516789,
+      "eval_rougeL": 0.531393156301234,
+      "eval_rougeLsum": 0.6434849921119885,
+      "eval_runtime": 15.3495,
+      "eval_samples_per_second": 0.651,
+      "eval_steps_per_second": 0.326,
+      "step": 26000
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4463755190372467,
+      "learning_rate": 5.0770501609365654e-05,
+      "loss": 1.4229,
+      "step": 26200
+    },
+    {
+      "epoch": 2.98,
+      "eval_bertscore": 0.7320259213447571,
+      "eval_loss": 1.7612769603729248,
+      "eval_rouge1": 0.6614975676412205,
+      "eval_rouge2": 0.3604125920722135,
+      "eval_rougeL": 0.5363549802785277,
+      "eval_rougeLsum": 0.6515851657527423,
+      "eval_runtime": 14.8919,
+      "eval_samples_per_second": 0.672,
+      "eval_steps_per_second": 0.336,
+      "step": 26200
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.40796974301338196,
+      "learning_rate": 4.963112769533142e-05,
+      "loss": 1.3813,
+      "step": 26400
+    },
+    {
+      "epoch": 3.01,
+      "eval_bertscore": 0.7295970320701599,
+      "eval_loss": 1.811266303062439,
+      "eval_rouge1": 0.6560490318552642,
+      "eval_rouge2": 0.3557119399635176,
+      "eval_rougeL": 0.5287962454708006,
+      "eval_rougeLsum": 0.6446835859698083,
+      "eval_runtime": 14.9171,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 26400
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.46325549483299255,
+      "learning_rate": 4.849175378129718e-05,
+      "loss": 1.2736,
+      "step": 26600
+    },
+    {
+      "epoch": 3.03,
+      "eval_bertscore": 0.7268515229225159,
+      "eval_loss": 1.8191944360733032,
+      "eval_rouge1": 0.6497892244364525,
+      "eval_rouge2": 0.3506188166717437,
+      "eval_rougeL": 0.5231517907924512,
+      "eval_rougeLsum": 0.637288999184507,
+      "eval_runtime": 15.1326,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 26600
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.418163925409317,
+      "learning_rate": 4.735237986726294e-05,
+      "loss": 1.2907,
+      "step": 26800
+    },
+    {
+      "epoch": 3.05,
+      "eval_bertscore": 0.7263709306716919,
+      "eval_loss": 1.8185272216796875,
+      "eval_rouge1": 0.6528251185371499,
+      "eval_rouge2": 0.35135402870682864,
+      "eval_rougeL": 0.5245953667859045,
+      "eval_rougeLsum": 0.6409906452290423,
+      "eval_runtime": 14.9485,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 26800
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.4502899944782257,
+      "learning_rate": 4.621300595322871e-05,
+      "loss": 1.312,
+      "step": 27000
+    },
+    {
+      "epoch": 3.08,
+      "eval_bertscore": 0.730769157409668,
+      "eval_loss": 1.8249473571777344,
+      "eval_rouge1": 0.6561686350412601,
+      "eval_rouge2": 0.3569324008900834,
+      "eval_rougeL": 0.5288774802109955,
+      "eval_rougeLsum": 0.6442271954437371,
+      "eval_runtime": 15.1193,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 27000
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.5041025280952454,
+      "learning_rate": 4.507363203919447e-05,
+      "loss": 1.2901,
+      "step": 27200
+    },
+    {
+      "epoch": 3.1,
+      "eval_bertscore": 0.7287546396255493,
+      "eval_loss": 1.827923059463501,
+      "eval_rouge1": 0.654431124870002,
+      "eval_rouge2": 0.35368874998515576,
+      "eval_rougeL": 0.5236566728859537,
+      "eval_rougeLsum": 0.641190956348846,
+      "eval_runtime": 15.038,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 27200
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.49700435996055603,
+      "learning_rate": 4.3934258125160227e-05,
+      "loss": 1.3022,
+      "step": 27400
+    },
+    {
+      "epoch": 3.12,
+      "eval_bertscore": 0.7300220727920532,
+      "eval_loss": 1.8278529644012451,
+      "eval_rouge1": 0.6495055712132203,
+      "eval_rouge2": 0.35008948755256797,
+      "eval_rougeL": 0.5213069893643358,
+      "eval_rougeLsum": 0.6390055765751734,
+      "eval_runtime": 15.0722,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 27400
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.5290235280990601,
+      "learning_rate": 4.279488421112599e-05,
+      "loss": 1.294,
+      "step": 27600
+    },
+    {
+      "epoch": 3.14,
+      "eval_bertscore": 0.729856014251709,
+      "eval_loss": 1.8240941762924194,
+      "eval_rouge1": 0.6470451896722034,
+      "eval_rouge2": 0.3467642899944511,
+      "eval_rougeL": 0.5174082021807023,
+      "eval_rougeLsum": 0.6342977220765258,
+      "eval_runtime": 15.0045,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 27600
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.48079508543014526,
+      "learning_rate": 4.1655510297091746e-05,
+      "loss": 1.306,
+      "step": 27800
+    },
+    {
+      "epoch": 3.17,
+      "eval_bertscore": 0.7321656346321106,
+      "eval_loss": 1.825411081314087,
+      "eval_rouge1": 0.6470417242893683,
+      "eval_rouge2": 0.34887648488076,
+      "eval_rougeL": 0.5195978025683929,
+      "eval_rougeLsum": 0.6352241217746025,
+      "eval_runtime": 15.2265,
+      "eval_samples_per_second": 0.657,
+      "eval_steps_per_second": 0.328,
+      "step": 27800
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.4453580379486084,
+      "learning_rate": 4.051613638305751e-05,
+      "loss": 1.2872,
+      "step": 28000
+    },
+    {
+      "epoch": 3.19,
+      "eval_bertscore": 0.7312643527984619,
+      "eval_loss": 1.8250818252563477,
+      "eval_rouge1": 0.6529323387086918,
+      "eval_rouge2": 0.35175299130000515,
+      "eval_rougeL": 0.5236016220703108,
+      "eval_rougeLsum": 0.641817087066691,
+      "eval_runtime": 15.2328,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 28000
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 0.5724875926971436,
+      "learning_rate": 3.937676246902327e-05,
+      "loss": 1.3023,
+      "step": 28200
+    },
+    {
+      "epoch": 3.21,
+      "eval_bertscore": 0.7281081080436707,
+      "eval_loss": 1.8279011249542236,
+      "eval_rouge1": 0.6513629794596312,
+      "eval_rouge2": 0.3514275809731242,
+      "eval_rougeL": 0.5223236211798525,
+      "eval_rougeLsum": 0.6403569684779228,
+      "eval_runtime": 15.1002,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28200
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.5111600756645203,
+      "learning_rate": 3.823738855498903e-05,
+      "loss": 1.3147,
+      "step": 28400
+    },
+    {
+      "epoch": 3.24,
+      "eval_bertscore": 0.7288422584533691,
+      "eval_loss": 1.8290317058563232,
+      "eval_rouge1": 0.6507807205572858,
+      "eval_rouge2": 0.347161808566521,
+      "eval_rougeL": 0.5183864656438575,
+      "eval_rougeLsum": 0.6389967703327324,
+      "eval_runtime": 15.1074,
+      "eval_samples_per_second": 0.662,
+      "eval_steps_per_second": 0.331,
+      "step": 28400
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.4945109486579895,
+      "learning_rate": 3.70980146409548e-05,
+      "loss": 1.2798,
+      "step": 28600
+    },
+    {
+      "epoch": 3.26,
+      "eval_bertscore": 0.7282342910766602,
+      "eval_loss": 1.8240516185760498,
+      "eval_rouge1": 0.6466411393254512,
+      "eval_rouge2": 0.3439998647318193,
+      "eval_rougeL": 0.5156993049024075,
+      "eval_rougeLsum": 0.6339007584931882,
+      "eval_runtime": 15.1238,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.331,
+      "step": 28600
+    },
+    {
+      "epoch": 3.28,
+      "grad_norm": 0.5064042806625366,
+      "learning_rate": 3.595864072692056e-05,
+      "loss": 1.2908,
+      "step": 28800
+    },
+    {
+      "epoch": 3.28,
+      "eval_bertscore": 0.7303023934364319,
+      "eval_loss": 1.8296499252319336,
+      "eval_rouge1": 0.6422335462244783,
+      "eval_rouge2": 0.3444891262545964,
+      "eval_rougeL": 0.5139669252888657,
+      "eval_rougeLsum": 0.6282864227941442,
+      "eval_runtime": 15.1801,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 28800
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.5009918808937073,
+      "learning_rate": 3.481926681288632e-05,
+      "loss": 1.2939,
+      "step": 29000
+    },
+    {
+      "epoch": 3.3,
+      "eval_bertscore": 0.728093147277832,
+      "eval_loss": 1.8265445232391357,
+      "eval_rouge1": 0.644883485667703,
+      "eval_rouge2": 0.3478659306090649,
+      "eval_rougeL": 0.5177213919602106,
+      "eval_rougeLsum": 0.6333667556178277,
+      "eval_runtime": 15.6279,
+      "eval_samples_per_second": 0.64,
+      "eval_steps_per_second": 0.32,
+      "step": 29000
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.5137434005737305,
+      "learning_rate": 3.3679892898852086e-05,
+      "loss": 1.2859,
+      "step": 29200
+    },
+    {
+      "epoch": 3.33,
+      "eval_bertscore": 0.7286854982376099,
+      "eval_loss": 1.8274028301239014,
+      "eval_rouge1": 0.6464900426605129,
+      "eval_rouge2": 0.34794140309215504,
+      "eval_rougeL": 0.5171376759249153,
+      "eval_rougeLsum": 0.6350675668052526,
+      "eval_runtime": 16.3341,
+      "eval_samples_per_second": 0.612,
+      "eval_steps_per_second": 0.306,
+      "step": 29200
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.5424513816833496,
+      "learning_rate": 3.254051898481784e-05,
+      "loss": 1.2982,
+      "step": 29400
+    },
+    {
+      "epoch": 3.35,
+      "eval_bertscore": 0.7272428274154663,
+      "eval_loss": 1.8286670446395874,
+      "eval_rouge1": 0.6481295498889761,
+      "eval_rouge2": 0.3477423613736981,
+      "eval_rougeL": 0.5196844939850981,
+      "eval_rougeLsum": 0.6363855713968847,
+      "eval_runtime": 15.5655,
+      "eval_samples_per_second": 0.642,
+      "eval_steps_per_second": 0.321,
+      "step": 29400
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.444140762090683,
+      "learning_rate": 3.1401145070783606e-05,
+      "loss": 1.31,
+      "step": 29600
+    },
+    {
+      "epoch": 3.37,
+      "eval_bertscore": 0.7301056981086731,
+      "eval_loss": 1.8282674551010132,
+      "eval_rouge1": 0.6540262039652377,
+      "eval_rouge2": 0.3482307679056835,
+      "eval_rougeL": 0.5226038858975284,
+      "eval_rougeLsum": 0.6417997985279889,
+      "eval_runtime": 15.2375,
+      "eval_samples_per_second": 0.656,
+      "eval_steps_per_second": 0.328,
+      "step": 29600
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.5904114842414856,
+      "learning_rate": 3.026177115674937e-05,
+      "loss": 1.294,
+      "step": 29800
+    },
+    {
+      "epoch": 3.39,
+      "eval_bertscore": 0.7272050976753235,
+      "eval_loss": 1.8291629552841187,
+      "eval_rouge1": 0.6451593111208989,
+      "eval_rouge2": 0.3441493766178606,
+      "eval_rougeL": 0.5159603215943032,
+      "eval_rougeLsum": 0.633764929096605,
+      "eval_runtime": 15.414,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 29800
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.4698268473148346,
+      "learning_rate": 2.9122397242715125e-05,
+      "loss": 1.302,
+      "step": 30000
+    },
+    {
+      "epoch": 3.42,
+      "eval_bertscore": 0.727090060710907,
+      "eval_loss": 1.826228141784668,
+      "eval_rouge1": 0.6437945042471859,
+      "eval_rouge2": 0.3459081721119077,
+      "eval_rougeL": 0.516648955547528,
+      "eval_rougeLsum": 0.6328083779823412,
+      "eval_runtime": 15.0752,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 30000
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.5351114869117737,
+      "learning_rate": 2.7983023328680892e-05,
+      "loss": 1.2971,
+      "step": 30200
+    },
+    {
+      "epoch": 3.44,
+      "eval_bertscore": 0.7274346947669983,
+      "eval_loss": 1.8284015655517578,
+      "eval_rouge1": 0.6428862858184051,
+      "eval_rouge2": 0.34402597597356294,
+      "eval_rougeL": 0.5169219417216804,
+      "eval_rougeLsum": 0.6311059754170927,
+      "eval_runtime": 15.1782,
+      "eval_samples_per_second": 0.659,
+      "eval_steps_per_second": 0.329,
+      "step": 30200
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.4645843207836151,
+      "learning_rate": 2.6843649414646655e-05,
+      "loss": 1.3118,
+      "step": 30400
+    },
+    {
+      "epoch": 3.46,
+      "eval_bertscore": 0.729422926902771,
+      "eval_loss": 1.8232523202896118,
+      "eval_rouge1": 0.6423334204787501,
+      "eval_rouge2": 0.34656408059281585,
+      "eval_rougeL": 0.5188486939610837,
+      "eval_rougeLsum": 0.631364621415121,
+      "eval_runtime": 15.069,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 30400
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.47329181432724,
+      "learning_rate": 2.5704275500612412e-05,
+      "loss": 1.2988,
+      "step": 30600
+    },
+    {
+      "epoch": 3.49,
+      "eval_bertscore": 0.7293562889099121,
+      "eval_loss": 1.828523874282837,
+      "eval_rouge1": 0.6415836961511305,
+      "eval_rouge2": 0.3448380376898492,
+      "eval_rougeL": 0.5160968947152185,
+      "eval_rougeLsum": 0.6307043736862319,
+      "eval_runtime": 15.0452,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.332,
+      "step": 30600
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.5455173850059509,
+      "learning_rate": 2.4564901586578175e-05,
+      "loss": 1.3047,
+      "step": 30800
+    },
+    {
+      "epoch": 3.51,
+      "eval_bertscore": 0.7271963357925415,
+      "eval_loss": 1.8309329748153687,
+      "eval_rouge1": 0.6426011693074479,
+      "eval_rouge2": 0.3431784961885528,
+      "eval_rougeL": 0.5149548951072465,
+      "eval_rougeLsum": 0.6316192615509898,
+      "eval_runtime": 15.6627,
+      "eval_samples_per_second": 0.638,
+      "eval_steps_per_second": 0.319,
+      "step": 30800
+    },
+    {
+      "epoch": 3.53,
+      "grad_norm": 0.5494771599769592,
+      "learning_rate": 2.342552767254394e-05,
+      "loss": 1.3024,
+      "step": 31000
+    },
+    {
+      "epoch": 3.53,
+      "eval_bertscore": 0.7272982597351074,
+      "eval_loss": 1.8302767276763916,
+      "eval_rouge1": 0.6466094927962344,
+      "eval_rouge2": 0.3471361046759306,
+      "eval_rougeL": 0.5189286897798941,
+      "eval_rougeLsum": 0.6353196842160578,
+      "eval_runtime": 15.1338,
+      "eval_samples_per_second": 0.661,
+      "eval_steps_per_second": 0.33,
+      "step": 31000
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 0.5403386354446411,
+      "learning_rate": 2.22861537585097e-05,
+      "loss": 1.297,
+      "step": 31200
+    },
+    {
+      "epoch": 3.55,
+      "eval_bertscore": 0.7270491123199463,
+      "eval_loss": 1.8272756338119507,
+      "eval_rouge1": 0.6403763553288392,
+      "eval_rouge2": 0.34466660447048997,
+      "eval_rougeL": 0.5148866492867579,
+      "eval_rougeLsum": 0.6292478620902813,
+      "eval_runtime": 14.964,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31200
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.503570020198822,
+      "learning_rate": 2.114677984447546e-05,
+      "loss": 1.2938,
+      "step": 31400
+    },
+    {
+      "epoch": 3.58,
+      "eval_bertscore": 0.7284318804740906,
+      "eval_loss": 1.829698920249939,
+      "eval_rouge1": 0.645157270000772,
+      "eval_rouge2": 0.3464589548126101,
+      "eval_rougeL": 0.520190218446654,
+      "eval_rougeLsum": 0.6339678909086379,
+      "eval_runtime": 14.9783,
+      "eval_samples_per_second": 0.668,
+      "eval_steps_per_second": 0.334,
+      "step": 31400
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.4921557307243347,
+      "learning_rate": 2.000740593044122e-05,
+      "loss": 1.2976,
+      "step": 31600
+    },
+    {
+      "epoch": 3.6,
+      "eval_bertscore": 0.7275441884994507,
+      "eval_loss": 1.826766014099121,
+      "eval_rouge1": 0.6448512195338201,
+      "eval_rouge2": 0.34423006917554333,
+      "eval_rougeL": 0.5158574771641147,
+      "eval_rougeLsum": 0.6334922214996468,
+      "eval_runtime": 15.0137,
+      "eval_samples_per_second": 0.666,
+      "eval_steps_per_second": 0.333,
+      "step": 31600
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 0.4597444534301758,
+      "learning_rate": 1.8868032016406985e-05,
+      "loss": 1.2984,
+      "step": 31800
+    },
+    {
+      "epoch": 3.62,
+      "eval_bertscore": 0.7296847105026245,
+      "eval_loss": 1.831121802330017,
+      "eval_rouge1": 0.6418710105695564,
+      "eval_rouge2": 0.34594102353175615,
+      "eval_rougeL": 0.5177915012238398,
+      "eval_rougeLsum": 0.6316847551673457,
+      "eval_runtime": 15.411,
+      "eval_samples_per_second": 0.649,
+      "eval_steps_per_second": 0.324,
+      "step": 31800
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.47503286600112915,
+      "learning_rate": 1.7728658102372748e-05,
+      "loss": 1.3009,
+      "step": 32000
+    },
+    {
+      "epoch": 3.65,
+      "eval_bertscore": 0.7298310995101929,
+      "eval_loss": 1.8273332118988037,
+      "eval_rouge1": 0.6417379366459213,
+      "eval_rouge2": 0.34542487508092823,
+      "eval_rougeL": 0.5158503557549705,
+      "eval_rougeLsum": 0.6309833082960914,
+      "eval_runtime": 14.9311,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 32000
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.45725110173225403,
+      "learning_rate": 1.6589284188338508e-05,
+      "loss": 1.3042,
+      "step": 32200
+    },
+    {
+      "epoch": 3.67,
+      "eval_bertscore": 0.7283689379692078,
+      "eval_loss": 1.8278729915618896,
+      "eval_rouge1": 0.6438134996619729,
+      "eval_rouge2": 0.34631038919793855,
+      "eval_rougeL": 0.5185962520496568,
+      "eval_rougeLsum": 0.6335181171392547,
+      "eval_runtime": 14.9577,
+      "eval_samples_per_second": 0.669,
+      "eval_steps_per_second": 0.334,
+      "step": 32200
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.527599036693573,
+      "learning_rate": 1.544991027430427e-05,
+      "loss": 1.3141,
+      "step": 32400
+    },
+    {
+      "epoch": 3.69,
+      "eval_bertscore": 0.7283642292022705,
+      "eval_loss": 1.828185796737671,
+      "eval_rouge1": 0.6420673962616871,
+      "eval_rouge2": 0.3428100609192179,
+      "eval_rougeL": 0.5151424402980586,
+      "eval_rougeLsum": 0.6312247707173357,
+      "eval_runtime": 15.0727,
+      "eval_samples_per_second": 0.663,
+      "eval_steps_per_second": 0.332,
+      "step": 32400
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.4286324679851532,
+      "learning_rate": 1.4310536360270033e-05,
+      "loss": 1.2965,
+      "step": 32600
+    },
+    {
+      "epoch": 3.71,
+      "eval_bertscore": 0.7269189953804016,
+      "eval_loss": 1.827340841293335,
+      "eval_rouge1": 0.6423441986765093,
+      "eval_rouge2": 0.3444233896929453,
+      "eval_rougeL": 0.515230719711361,
+      "eval_rougeLsum": 0.6324502593091261,
+      "eval_runtime": 15.762,
+      "eval_samples_per_second": 0.634,
+      "eval_steps_per_second": 0.317,
+      "step": 32600
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.45857399702072144,
+      "learning_rate": 1.3171162446235794e-05,
+      "loss": 1.3106,
+      "step": 32800
+    },
+    {
+      "epoch": 3.74,
+      "eval_bertscore": 0.7266061305999756,
+      "eval_loss": 1.8283151388168335,
+      "eval_rouge1": 0.6411621669658236,
+      "eval_rouge2": 0.3443038317384875,
+      "eval_rougeL": 0.5144500926275494,
+      "eval_rougeLsum": 0.6309220735111227,
+      "eval_runtime": 15.7907,
+      "eval_samples_per_second": 0.633,
+      "eval_steps_per_second": 0.317,
+      "step": 32800
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.5014874935150146,
+      "learning_rate": 1.2031788532201556e-05,
+      "loss": 1.2949,
+      "step": 33000
+    },
+    {
+      "epoch": 3.76,
+      "eval_bertscore": 0.7257188558578491,
+      "eval_loss": 1.8288466930389404,
+      "eval_rouge1": 0.6423913963160501,
+      "eval_rouge2": 0.3441299434811822,
+      "eval_rougeL": 0.5144996687729989,
+      "eval_rougeLsum": 0.6320056022872689,
+      "eval_runtime": 15.28,
+      "eval_samples_per_second": 0.654,
+      "eval_steps_per_second": 0.327,
+      "step": 33000
+    },
+    {
+      "epoch": 3.78,
+      "grad_norm": 0.45499473810195923,
+      "learning_rate": 1.0892414618167317e-05,
+      "loss": 1.3217,
+      "step": 33200
+    },
+    {
+      "epoch": 3.78,
+      "eval_bertscore": 0.7278145551681519,
+      "eval_loss": 1.8302255868911743,
+      "eval_rouge1": 0.6425439428330799,
+      "eval_rouge2": 0.34625777040537886,
+      "eval_rougeL": 0.5154002067201906,
+      "eval_rougeLsum": 0.6311754713033721,
+      "eval_runtime": 14.9234,
+      "eval_samples_per_second": 0.67,
+      "eval_steps_per_second": 0.335,
+      "step": 33200
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.528462827205658,
+      "learning_rate": 9.753040704133079e-06,
+      "loss": 1.2944,
+      "step": 33400
+    },
+    {
+      "epoch": 3.8,
+      "eval_bertscore": 0.7256959676742554,
+      "eval_loss": 1.8285919427871704,
+      "eval_rouge1": 0.6427339109412973,
+      "eval_rouge2": 0.34490192490238136,
+      "eval_rougeL": 0.5158098895481885,
+      "eval_rougeLsum": 0.6318354230613757,
+      "eval_runtime": 14.821,
+      "eval_samples_per_second": 0.675,
+      "eval_steps_per_second": 0.337,
+      "step": 33400
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.4865368604660034,
+      "learning_rate": 8.61366679009884e-06,
+      "loss": 1.2974,
+      "step": 33600
+    },
+    {
+      "epoch": 3.83,
+      "eval_bertscore": 0.726431667804718,
+      "eval_loss": 1.827730417251587,
+      "eval_rouge1": 0.6411658367235875,
+      "eval_rouge2": 0.34535262367337394,
+      "eval_rougeL": 0.5145257214263663,
+      "eval_rougeLsum": 0.630130047089762,
+      "eval_runtime": 15.0627,
+      "eval_samples_per_second": 0.664,
+      "eval_steps_per_second": 0.332,
+      "step": 33600
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.5641516447067261,
+      "learning_rate": 7.474292876064602e-06,
+      "loss": 1.3205,
+      "step": 33800
+    },
+    {
+      "epoch": 3.85,
+      "eval_bertscore": 0.7264658808708191,
+      "eval_loss": 1.8282066583633423,
+      "eval_rouge1": 0.6453450156795362,
+      "eval_rouge2": 0.34482901597860427,
+      "eval_rougeL": 0.5163102729864533,
+      "eval_rougeLsum": 0.6341114211680331,
+      "eval_runtime": 15.0375,
+      "eval_samples_per_second": 0.665,
+      "eval_steps_per_second": 0.333,
+      "step": 33800
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.4884029030799866,
+      "learning_rate": 6.3349189620303645e-06,
+      "loss": 1.2843,
+      "step": 34000
+    },
+    {
+      "epoch": 3.87,
+      "eval_bertscore": 0.727534294128418,
+      "eval_loss": 1.826780915260315,
+      "eval_rouge1": 0.6473030907072772,
+      "eval_rouge2": 0.3478476140487603,
+      "eval_rougeL": 0.5181912668392612,
+      "eval_rougeLsum": 0.6360238739725694,
+      "eval_runtime": 15.5284,
+      "eval_samples_per_second": 0.644,
+      "eval_steps_per_second": 0.322,
+      "step": 34000
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.5166618824005127,
+      "learning_rate": 5.195545047996126e-06,
+      "loss": 1.3082,
+      "step": 34200
+    },
+    {
+      "epoch": 3.9,
+      "eval_bertscore": 0.7270146608352661,
+      "eval_loss": 1.8250738382339478,
+      "eval_rouge1": 0.6437340233509039,
+      "eval_rouge2": 0.34801822195059,
+      "eval_rougeL": 0.517041673366943,
+      "eval_rougeLsum": 0.6322916261036109,
+      "eval_runtime": 15.8529,
+      "eval_samples_per_second": 0.631,
+      "eval_steps_per_second": 0.315,
+      "step": 34200
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.4907975196838379,
+      "learning_rate": 4.0561711339618885e-06,
+      "loss": 1.2885,
+      "step": 34400
+    },
+    {
+      "epoch": 3.92,
+      "eval_bertscore": 0.7279630303382874,
+      "eval_loss": 1.825722336769104,
+      "eval_rouge1": 0.6460412710248096,
+      "eval_rouge2": 0.3472000197615478,
+      "eval_rougeL": 0.5184161078282452,
+      "eval_rougeLsum": 0.6354684688419843,
+      "eval_runtime": 14.9112,
+      "eval_samples_per_second": 0.671,
+      "eval_steps_per_second": 0.335,
+      "step": 34400
+    }
+  ],
+  "logging_steps": 200,
+  "max_steps": 35112,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 800,
+  "total_flos": 3.483911335305044e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-34400/training_args.bin b/checkpoint-34400/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-34400/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048
diff --git a/checkpoint-epoch-4/README.md b/checkpoint-epoch-4/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..06c7596b5c172681580254734d25e13589eecea4
--- /dev/null
+++ b/checkpoint-epoch-4/README.md
@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: unsloth/gemma-2b-bnb-4bit
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.10.0
\ No newline at end of file
diff --git a/checkpoint-epoch-4/adapter_config.json b/checkpoint-epoch-4/adapter_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..4a4fe22f6834f8822b37529e002cb87176abb2da
--- /dev/null
+++ b/checkpoint-epoch-4/adapter_config.json
@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "unsloth/gemma-2b-bnb-4bit",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_dropout": 0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": "unsloth",
+  "target_modules": [
+    "v_proj",
+    "up_proj",
+    "gate_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}
\ No newline at end of file
diff --git a/checkpoint-epoch-4/adapter_model.safetensors b/checkpoint-epoch-4/adapter_model.safetensors
new file mode 100644
index 0000000000000000000000000000000000000000..704ba965999c74513494b4acf484683aaabf85a0
--- /dev/null
+++ b/checkpoint-epoch-4/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f6a8ed5c5dc49678cb4883e3b329811fecffd783aa9d1a9ce8d90981d875b94
+size 313820248
diff --git a/checkpoint-epoch-4/special_tokens_map.json b/checkpoint-epoch-4/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..8d6368f7e735fbe4781bf6e956b7c6ad0586df80
--- /dev/null
+++ b/checkpoint-epoch-4/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-epoch-4/tokenizer.json b/checkpoint-epoch-4/tokenizer.json
new file mode 100644
index 0000000000000000000000000000000000000000..10d1ca087d5e78cbfc41ae083201df86aa031044
--- /dev/null
+++ b/checkpoint-epoch-4/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05e97791a5e007260de1db7e1692e53150e08cea481e2bf25435553380c147ee
+size 17477929
diff --git a/checkpoint-epoch-4/tokenizer.model b/checkpoint-epoch-4/tokenizer.model
new file mode 100644
index 0000000000000000000000000000000000000000..796efe9ab515c15e146ce7588e6d7b9b8134dbf8
--- /dev/null
+++ b/checkpoint-epoch-4/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-epoch-4/tokenizer_config.json b/checkpoint-epoch-4/tokenizer_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..9333d18c56676f1d27051f3207b4d13c00df1684
--- /dev/null
+++ b/checkpoint-epoch-4/tokenizer_config.json
@@ -0,0 +1,71 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "chat_template": "{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '\n' + message['content'] | trim + '<end_of_turn>\n' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model\n'}}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "legacy": null,
+  "model_max_length": 8192,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-epoch-4/training_args.bin b/checkpoint-epoch-4/training_args.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2539ba5804bc6588934ce4ed9e2791965ab96e71
--- /dev/null
+++ b/checkpoint-epoch-4/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d576743d4294cfbcae212243377437f73cabbcded2df9da7f73e6e39ecb38f56
+size 5048