Commit
•
0525ea8
1
Parent(s):
c3abef6
Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +7 -0
- train_results.json +9 -0
- trainer_state.json +522 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: tinyllama-1.1b-chat-v1.0-hh-dpo
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# tinyllama-1.1b-chat-v1.0-hh-dpo
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on an unknown dataset.
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 5e-07
|
| 39 |
+
- train_batch_size: 16
|
| 40 |
+
- eval_batch_size: 16
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- num_devices: 4
|
| 44 |
+
- gradient_accumulation_steps: 8
|
| 45 |
+
- total_train_batch_size: 512
|
| 46 |
+
- total_eval_batch_size: 64
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: cosine
|
| 49 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 1
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- Transformers 4.44.1
|
| 59 |
+
- Pytorch 2.1.2+cu121
|
| 60 |
+
- Datasets 2.21.0
|
| 61 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996020692399522,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.678918091354856,
|
| 5 |
+
"train_runtime": 7383.0158,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 21.78,
|
| 8 |
+
"train_steps_per_second": 0.043
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"max_length": 2048,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.44.1"
|
| 7 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996020692399522,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.678918091354856,
|
| 5 |
+
"train_runtime": 7383.0158,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 21.78,
|
| 8 |
+
"train_steps_per_second": 0.043
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,522 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9996020692399522,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 314,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0031834460803820135,
|
| 13 |
+
"grad_norm": 0.8913188458215807,
|
| 14 |
+
"learning_rate": 1.5625e-08,
|
| 15 |
+
"logits/chosen": -2.332231044769287,
|
| 16 |
+
"logits/rejected": -2.3125171661376953,
|
| 17 |
+
"logps/chosen": -178.02963256835938,
|
| 18 |
+
"logps/rejected": -150.3365478515625,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.03183446080382014,
|
| 28 |
+
"grad_norm": 0.7954378994525454,
|
| 29 |
+
"learning_rate": 1.5624999999999999e-07,
|
| 30 |
+
"logits/chosen": -2.299248456954956,
|
| 31 |
+
"logits/rejected": -2.2925186157226562,
|
| 32 |
+
"logps/chosen": -160.2787322998047,
|
| 33 |
+
"logps/rejected": -158.51319885253906,
|
| 34 |
+
"loss": 0.6931,
|
| 35 |
+
"rewards/accuracies": 0.4322916567325592,
|
| 36 |
+
"rewards/chosen": -3.1086481612874195e-05,
|
| 37 |
+
"rewards/margins": -0.00011221379099879414,
|
| 38 |
+
"rewards/rejected": 8.112730574794114e-05,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06366892160764027,
|
| 43 |
+
"grad_norm": 0.856612493875923,
|
| 44 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 45 |
+
"logits/chosen": -2.2908735275268555,
|
| 46 |
+
"logits/rejected": -2.294975519180298,
|
| 47 |
+
"logps/chosen": -156.20362854003906,
|
| 48 |
+
"logps/rejected": -160.60763549804688,
|
| 49 |
+
"loss": 0.6932,
|
| 50 |
+
"rewards/accuracies": 0.48828125,
|
| 51 |
+
"rewards/chosen": -9.316079376731068e-05,
|
| 52 |
+
"rewards/margins": 9.743528607941698e-06,
|
| 53 |
+
"rewards/rejected": -0.00010290431964676827,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.0955033824114604,
|
| 58 |
+
"grad_norm": 0.8601850412460429,
|
| 59 |
+
"learning_rate": 4.6874999999999996e-07,
|
| 60 |
+
"logits/chosen": -2.2940285205841064,
|
| 61 |
+
"logits/rejected": -2.2888407707214355,
|
| 62 |
+
"logps/chosen": -159.58680725097656,
|
| 63 |
+
"logps/rejected": -162.42189025878906,
|
| 64 |
+
"loss": 0.693,
|
| 65 |
+
"rewards/accuracies": 0.52734375,
|
| 66 |
+
"rewards/chosen": -0.0016975710168480873,
|
| 67 |
+
"rewards/margins": 0.00033986102789640427,
|
| 68 |
+
"rewards/rejected": -0.0020374320447444916,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.12733784321528055,
|
| 73 |
+
"grad_norm": 0.8744440488120392,
|
| 74 |
+
"learning_rate": 4.990077890125363e-07,
|
| 75 |
+
"logits/chosen": -2.301192045211792,
|
| 76 |
+
"logits/rejected": -2.2977750301361084,
|
| 77 |
+
"logps/chosen": -151.9224395751953,
|
| 78 |
+
"logps/rejected": -152.43655395507812,
|
| 79 |
+
"loss": 0.6925,
|
| 80 |
+
"rewards/accuracies": 0.555468738079071,
|
| 81 |
+
"rewards/chosen": -0.005973272956907749,
|
| 82 |
+
"rewards/margins": 0.0012832467909902334,
|
| 83 |
+
"rewards/rejected": -0.007256519980728626,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.15917230401910068,
|
| 88 |
+
"grad_norm": 0.95273331836508,
|
| 89 |
+
"learning_rate": 4.949904262591467e-07,
|
| 90 |
+
"logits/chosen": -2.3083178997039795,
|
| 91 |
+
"logits/rejected": -2.30576753616333,
|
| 92 |
+
"logps/chosen": -161.62750244140625,
|
| 93 |
+
"logps/rejected": -159.9287109375,
|
| 94 |
+
"loss": 0.6914,
|
| 95 |
+
"rewards/accuracies": 0.534375011920929,
|
| 96 |
+
"rewards/chosen": -0.016185810789465904,
|
| 97 |
+
"rewards/margins": 0.003111806232482195,
|
| 98 |
+
"rewards/rejected": -0.019297616556286812,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.1910067648229208,
|
| 103 |
+
"grad_norm": 0.9051345997712666,
|
| 104 |
+
"learning_rate": 4.879356673988089e-07,
|
| 105 |
+
"logits/chosen": -2.3459715843200684,
|
| 106 |
+
"logits/rejected": -2.340350389480591,
|
| 107 |
+
"logps/chosen": -159.88613891601562,
|
| 108 |
+
"logps/rejected": -162.18612670898438,
|
| 109 |
+
"loss": 0.6903,
|
| 110 |
+
"rewards/accuracies": 0.555468738079071,
|
| 111 |
+
"rewards/chosen": -0.029835861176252365,
|
| 112 |
+
"rewards/margins": 0.005405827891081572,
|
| 113 |
+
"rewards/rejected": -0.03524169698357582,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.22284122562674094,
|
| 118 |
+
"grad_norm": 0.9611988462128772,
|
| 119 |
+
"learning_rate": 4.779309774701573e-07,
|
| 120 |
+
"logits/chosen": -2.3523900508880615,
|
| 121 |
+
"logits/rejected": -2.348705768585205,
|
| 122 |
+
"logps/chosen": -166.91146850585938,
|
| 123 |
+
"logps/rejected": -164.3793182373047,
|
| 124 |
+
"loss": 0.6883,
|
| 125 |
+
"rewards/accuracies": 0.5609375238418579,
|
| 126 |
+
"rewards/chosen": -0.05225539207458496,
|
| 127 |
+
"rewards/margins": 0.009335539303719997,
|
| 128 |
+
"rewards/rejected": -0.06159093230962753,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2546756864305611,
|
| 133 |
+
"grad_norm": 0.9747994880646563,
|
| 134 |
+
"learning_rate": 4.6510039481503485e-07,
|
| 135 |
+
"logits/chosen": -2.367791175842285,
|
| 136 |
+
"logits/rejected": -2.3638641834259033,
|
| 137 |
+
"logps/chosen": -166.5675811767578,
|
| 138 |
+
"logps/rejected": -163.23748779296875,
|
| 139 |
+
"loss": 0.6866,
|
| 140 |
+
"rewards/accuracies": 0.56640625,
|
| 141 |
+
"rewards/chosen": -0.0814916118979454,
|
| 142 |
+
"rewards/margins": 0.01518010813742876,
|
| 143 |
+
"rewards/rejected": -0.09667172282934189,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.28651014723438123,
|
| 148 |
+
"grad_norm": 1.0299054182265805,
|
| 149 |
+
"learning_rate": 4.4960299324869857e-07,
|
| 150 |
+
"logits/chosen": -2.3870062828063965,
|
| 151 |
+
"logits/rejected": -2.382286787033081,
|
| 152 |
+
"logps/chosen": -168.0864715576172,
|
| 153 |
+
"logps/rejected": -170.67047119140625,
|
| 154 |
+
"loss": 0.6857,
|
| 155 |
+
"rewards/accuracies": 0.5453125238418579,
|
| 156 |
+
"rewards/chosen": -0.10201652348041534,
|
| 157 |
+
"rewards/margins": 0.017032291740179062,
|
| 158 |
+
"rewards/rejected": -0.11904881149530411,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.31834460803820136,
|
| 163 |
+
"grad_norm": 1.0287716469740125,
|
| 164 |
+
"learning_rate": 4.3163090985954074e-07,
|
| 165 |
+
"logits/chosen": -2.3779287338256836,
|
| 166 |
+
"logits/rejected": -2.369999408721924,
|
| 167 |
+
"logps/chosen": -170.18251037597656,
|
| 168 |
+
"logps/rejected": -172.33718872070312,
|
| 169 |
+
"loss": 0.6838,
|
| 170 |
+
"rewards/accuracies": 0.5835937261581421,
|
| 171 |
+
"rewards/chosen": -0.12887540459632874,
|
| 172 |
+
"rewards/margins": 0.023395564407110214,
|
| 173 |
+
"rewards/rejected": -0.15227097272872925,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.3501790688420215,
|
| 178 |
+
"grad_norm": 1.0989482799498032,
|
| 179 |
+
"learning_rate": 4.114069628897006e-07,
|
| 180 |
+
"logits/chosen": -2.358673572540283,
|
| 181 |
+
"logits/rejected": -2.3572804927825928,
|
| 182 |
+
"logps/chosen": -173.09437561035156,
|
| 183 |
+
"logps/rejected": -176.59164428710938,
|
| 184 |
+
"loss": 0.681,
|
| 185 |
+
"rewards/accuracies": 0.5703125,
|
| 186 |
+
"rewards/chosen": -0.14945295453071594,
|
| 187 |
+
"rewards/margins": 0.03385554999113083,
|
| 188 |
+
"rewards/rejected": -0.18330851197242737,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.3820135296458416,
|
| 193 |
+
"grad_norm": 1.215724859704609,
|
| 194 |
+
"learning_rate": 3.891818892301304e-07,
|
| 195 |
+
"logits/chosen": -2.3581154346466064,
|
| 196 |
+
"logits/rejected": -2.3629350662231445,
|
| 197 |
+
"logps/chosen": -176.79432678222656,
|
| 198 |
+
"logps/rejected": -180.97039794921875,
|
| 199 |
+
"loss": 0.6816,
|
| 200 |
+
"rewards/accuracies": 0.563281238079071,
|
| 201 |
+
"rewards/chosen": -0.2229323834180832,
|
| 202 |
+
"rewards/margins": 0.02610206976532936,
|
| 203 |
+
"rewards/rejected": -0.24903444945812225,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.41384799044966175,
|
| 208 |
+
"grad_norm": 1.276372767011757,
|
| 209 |
+
"learning_rate": 3.6523123577970693e-07,
|
| 210 |
+
"logits/chosen": -2.362705707550049,
|
| 211 |
+
"logits/rejected": -2.3545660972595215,
|
| 212 |
+
"logps/chosen": -187.2711944580078,
|
| 213 |
+
"logps/rejected": -192.3023681640625,
|
| 214 |
+
"loss": 0.6774,
|
| 215 |
+
"rewards/accuracies": 0.577343761920929,
|
| 216 |
+
"rewards/chosen": -0.2734658420085907,
|
| 217 |
+
"rewards/margins": 0.04484738036990166,
|
| 218 |
+
"rewards/rejected": -0.31831321120262146,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.4456824512534819,
|
| 223 |
+
"grad_norm": 1.3874387536648047,
|
| 224 |
+
"learning_rate": 3.3985194320937815e-07,
|
| 225 |
+
"logits/chosen": -2.3698925971984863,
|
| 226 |
+
"logits/rejected": -2.375953435897827,
|
| 227 |
+
"logps/chosen": -189.7623748779297,
|
| 228 |
+
"logps/rejected": -198.46641540527344,
|
| 229 |
+
"loss": 0.6765,
|
| 230 |
+
"rewards/accuracies": 0.577343761920929,
|
| 231 |
+
"rewards/chosen": -0.28896641731262207,
|
| 232 |
+
"rewards/margins": 0.0472131185233593,
|
| 233 |
+
"rewards/rejected": -0.3361795246601105,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.477516912057302,
|
| 238 |
+
"grad_norm": 1.4443860790246479,
|
| 239 |
+
"learning_rate": 3.133586644859039e-07,
|
| 240 |
+
"logits/chosen": -2.3575439453125,
|
| 241 |
+
"logits/rejected": -2.3553812503814697,
|
| 242 |
+
"logps/chosen": -191.13748168945312,
|
| 243 |
+
"logps/rejected": -193.09817504882812,
|
| 244 |
+
"loss": 0.6777,
|
| 245 |
+
"rewards/accuracies": 0.5640624761581421,
|
| 246 |
+
"rewards/chosen": -0.3255929946899414,
|
| 247 |
+
"rewards/margins": 0.048027556389570236,
|
| 248 |
+
"rewards/rejected": -0.37362051010131836,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.5093513728611222,
|
| 253 |
+
"grad_norm": 1.6146506390035185,
|
| 254 |
+
"learning_rate": 2.8607986379820664e-07,
|
| 255 |
+
"logits/chosen": -2.3630471229553223,
|
| 256 |
+
"logits/rejected": -2.3578333854675293,
|
| 257 |
+
"logps/chosen": -203.55029296875,
|
| 258 |
+
"logps/rejected": -198.3260955810547,
|
| 259 |
+
"loss": 0.6749,
|
| 260 |
+
"rewards/accuracies": 0.543749988079071,
|
| 261 |
+
"rewards/chosen": -0.36638548970222473,
|
| 262 |
+
"rewards/margins": 0.03375838324427605,
|
| 263 |
+
"rewards/rejected": -0.4001438617706299,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.5411858336649423,
|
| 268 |
+
"grad_norm": 1.442156527952266,
|
| 269 |
+
"learning_rate": 2.583537442519186e-07,
|
| 270 |
+
"logits/chosen": -2.37186861038208,
|
| 271 |
+
"logits/rejected": -2.37400484085083,
|
| 272 |
+
"logps/chosen": -196.1448211669922,
|
| 273 |
+
"logps/rejected": -203.14988708496094,
|
| 274 |
+
"loss": 0.6733,
|
| 275 |
+
"rewards/accuracies": 0.571093738079071,
|
| 276 |
+
"rewards/chosen": -0.39402055740356445,
|
| 277 |
+
"rewards/margins": 0.05580953508615494,
|
| 278 |
+
"rewards/rejected": -0.4498301148414612,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.5730202944687625,
|
| 283 |
+
"grad_norm": 1.6421108977251095,
|
| 284 |
+
"learning_rate": 2.3052405482064919e-07,
|
| 285 |
+
"logits/chosen": -2.375570297241211,
|
| 286 |
+
"logits/rejected": -2.3756964206695557,
|
| 287 |
+
"logps/chosen": -210.69284057617188,
|
| 288 |
+
"logps/rejected": -210.54672241210938,
|
| 289 |
+
"loss": 0.6783,
|
| 290 |
+
"rewards/accuracies": 0.550000011920929,
|
| 291 |
+
"rewards/chosen": -0.4329681396484375,
|
| 292 |
+
"rewards/margins": 0.0399855375289917,
|
| 293 |
+
"rewards/rejected": -0.4729536473751068,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.6048547552725826,
|
| 298 |
+
"grad_norm": 1.8616819792733938,
|
| 299 |
+
"learning_rate": 2.029358285394716e-07,
|
| 300 |
+
"logits/chosen": -2.375413179397583,
|
| 301 |
+
"logits/rejected": -2.3708198070526123,
|
| 302 |
+
"logps/chosen": -206.62564086914062,
|
| 303 |
+
"logps/rejected": -206.83370971679688,
|
| 304 |
+
"loss": 0.6694,
|
| 305 |
+
"rewards/accuracies": 0.609375,
|
| 306 |
+
"rewards/chosen": -0.41667842864990234,
|
| 307 |
+
"rewards/margins": 0.06583230942487717,
|
| 308 |
+
"rewards/rejected": -0.4825107455253601,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.6366892160764027,
|
| 313 |
+
"grad_norm": 2.107824221269726,
|
| 314 |
+
"learning_rate": 1.7593110477859152e-07,
|
| 315 |
+
"logits/chosen": -2.3875057697296143,
|
| 316 |
+
"logits/rejected": -2.387460470199585,
|
| 317 |
+
"logps/chosen": -201.22967529296875,
|
| 318 |
+
"logps/rejected": -205.489501953125,
|
| 319 |
+
"loss": 0.6713,
|
| 320 |
+
"rewards/accuracies": 0.5757812261581421,
|
| 321 |
+
"rewards/chosen": -0.42719680070877075,
|
| 322 |
+
"rewards/margins": 0.06031234189867973,
|
| 323 |
+
"rewards/rejected": -0.487509161233902,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.6685236768802229,
|
| 328 |
+
"grad_norm": 2.4249761136425048,
|
| 329 |
+
"learning_rate": 1.4984468863253007e-07,
|
| 330 |
+
"logits/chosen": -2.394925594329834,
|
| 331 |
+
"logits/rejected": -2.386389970779419,
|
| 332 |
+
"logps/chosen": -215.95462036132812,
|
| 333 |
+
"logps/rejected": -215.75820922851562,
|
| 334 |
+
"loss": 0.6748,
|
| 335 |
+
"rewards/accuracies": 0.569531261920929,
|
| 336 |
+
"rewards/chosen": -0.4547385275363922,
|
| 337 |
+
"rewards/margins": 0.05981076508760452,
|
| 338 |
+
"rewards/rejected": -0.5145493149757385,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.700358137684043,
|
| 343 |
+
"grad_norm": 1.8116066543619882,
|
| 344 |
+
"learning_rate": 1.2500000000000005e-07,
|
| 345 |
+
"logits/chosen": -2.3963189125061035,
|
| 346 |
+
"logits/rejected": -2.3958048820495605,
|
| 347 |
+
"logps/chosen": -212.20693969726562,
|
| 348 |
+
"logps/rejected": -216.2605743408203,
|
| 349 |
+
"loss": 0.6721,
|
| 350 |
+
"rewards/accuracies": 0.5757812261581421,
|
| 351 |
+
"rewards/chosen": -0.4895060658454895,
|
| 352 |
+
"rewards/margins": 0.06058992072939873,
|
| 353 |
+
"rewards/rejected": -0.5500959157943726,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.7321925984878631,
|
| 358 |
+
"grad_norm": 1.8113200913829595,
|
| 359 |
+
"learning_rate": 1.0170506381766119e-07,
|
| 360 |
+
"logits/chosen": -2.387866497039795,
|
| 361 |
+
"logits/rejected": -2.388319730758667,
|
| 362 |
+
"logps/chosen": -207.97604370117188,
|
| 363 |
+
"logps/rejected": -218.859130859375,
|
| 364 |
+
"loss": 0.6722,
|
| 365 |
+
"rewards/accuracies": 0.586718738079071,
|
| 366 |
+
"rewards/chosen": -0.49019813537597656,
|
| 367 |
+
"rewards/margins": 0.07117091119289398,
|
| 368 |
+
"rewards/rejected": -0.5613690614700317,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.7640270592916832,
|
| 373 |
+
"grad_norm": 2.1202663184992563,
|
| 374 |
+
"learning_rate": 8.024869116091879e-08,
|
| 375 |
+
"logits/chosen": -2.3979618549346924,
|
| 376 |
+
"logits/rejected": -2.406341075897217,
|
| 377 |
+
"logps/chosen": -219.69271850585938,
|
| 378 |
+
"logps/rejected": -221.71139526367188,
|
| 379 |
+
"loss": 0.6734,
|
| 380 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 381 |
+
"rewards/chosen": -0.4954513907432556,
|
| 382 |
+
"rewards/margins": 0.05036097764968872,
|
| 383 |
+
"rewards/rejected": -0.5458123683929443,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7958615200955034,
|
| 388 |
+
"grad_norm": 1.8622859998372623,
|
| 389 |
+
"learning_rate": 6.089689855854869e-08,
|
| 390 |
+
"logits/chosen": -2.3931431770324707,
|
| 391 |
+
"logits/rejected": -2.4037561416625977,
|
| 392 |
+
"logps/chosen": -207.83798217773438,
|
| 393 |
+
"logps/rejected": -216.9298858642578,
|
| 394 |
+
"loss": 0.6728,
|
| 395 |
+
"rewards/accuracies": 0.5835937261581421,
|
| 396 |
+
"rewards/chosen": -0.4849010407924652,
|
| 397 |
+
"rewards/margins": 0.0694160908460617,
|
| 398 |
+
"rewards/rejected": -0.5543171167373657,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.8276959808993235,
|
| 403 |
+
"grad_norm": 1.8122633866114284,
|
| 404 |
+
"learning_rate": 4.388960991455998e-08,
|
| 405 |
+
"logits/chosen": -2.3899784088134766,
|
| 406 |
+
"logits/rejected": -2.395357847213745,
|
| 407 |
+
"logps/chosen": -205.0048828125,
|
| 408 |
+
"logps/rejected": -213.7442626953125,
|
| 409 |
+
"loss": 0.6718,
|
| 410 |
+
"rewards/accuracies": 0.5882812738418579,
|
| 411 |
+
"rewards/chosen": -0.4997900128364563,
|
| 412 |
+
"rewards/margins": 0.0625927522778511,
|
| 413 |
+
"rewards/rejected": -0.5623827576637268,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.8595304417031436,
|
| 418 |
+
"grad_norm": 2.0654372503377014,
|
| 419 |
+
"learning_rate": 2.943768192692958e-08,
|
| 420 |
+
"logits/chosen": -2.400757074356079,
|
| 421 |
+
"logits/rejected": -2.4067251682281494,
|
| 422 |
+
"logps/chosen": -205.0299072265625,
|
| 423 |
+
"logps/rejected": -216.3090362548828,
|
| 424 |
+
"loss": 0.6731,
|
| 425 |
+
"rewards/accuracies": 0.5640624761581421,
|
| 426 |
+
"rewards/chosen": -0.4922141134738922,
|
| 427 |
+
"rewards/margins": 0.06625890731811523,
|
| 428 |
+
"rewards/rejected": -0.5584729313850403,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.8913649025069638,
|
| 433 |
+
"grad_norm": 2.2123553368333644,
|
| 434 |
+
"learning_rate": 1.7720289882128092e-08,
|
| 435 |
+
"logits/chosen": -2.3984594345092773,
|
| 436 |
+
"logits/rejected": -2.4013664722442627,
|
| 437 |
+
"logps/chosen": -204.21585083007812,
|
| 438 |
+
"logps/rejected": -210.4906768798828,
|
| 439 |
+
"loss": 0.6745,
|
| 440 |
+
"rewards/accuracies": 0.5453125238418579,
|
| 441 |
+
"rewards/chosen": -0.497935950756073,
|
| 442 |
+
"rewards/margins": 0.04604557901620865,
|
| 443 |
+
"rewards/rejected": -0.5439816117286682,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.9231993633107839,
|
| 448 |
+
"grad_norm": 2.8178850748361763,
|
| 449 |
+
"learning_rate": 8.882706236405885e-09,
|
| 450 |
+
"logits/chosen": -2.3987581729888916,
|
| 451 |
+
"logits/rejected": -2.398507833480835,
|
| 452 |
+
"logps/chosen": -209.7099609375,
|
| 453 |
+
"logps/rejected": -216.7168426513672,
|
| 454 |
+
"loss": 0.6672,
|
| 455 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 456 |
+
"rewards/chosen": -0.49557456374168396,
|
| 457 |
+
"rewards/margins": 0.07922448217868805,
|
| 458 |
+
"rewards/rejected": -0.5747990608215332,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.955033824114604,
|
| 463 |
+
"grad_norm": 2.094778053644434,
|
| 464 |
+
"learning_rate": 3.0344995250326245e-09,
|
| 465 |
+
"logits/chosen": -2.4099345207214355,
|
| 466 |
+
"logits/rejected": -2.406144380569458,
|
| 467 |
+
"logps/chosen": -210.4052734375,
|
| 468 |
+
"logps/rejected": -220.04794311523438,
|
| 469 |
+
"loss": 0.6676,
|
| 470 |
+
"rewards/accuracies": 0.5921875238418579,
|
| 471 |
+
"rewards/chosen": -0.5059640407562256,
|
| 472 |
+
"rewards/margins": 0.0776122659444809,
|
| 473 |
+
"rewards/rejected": -0.5835763216018677,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.9868682849184242,
|
| 478 |
+
"grad_norm": 2.5156498514265735,
|
| 479 |
+
"learning_rate": 2.481759294498398e-10,
|
| 480 |
+
"logits/chosen": -2.3863139152526855,
|
| 481 |
+
"logits/rejected": -2.3877015113830566,
|
| 482 |
+
"logps/chosen": -207.8264923095703,
|
| 483 |
+
"logps/rejected": -215.2146453857422,
|
| 484 |
+
"loss": 0.6713,
|
| 485 |
+
"rewards/accuracies": 0.5679687261581421,
|
| 486 |
+
"rewards/chosen": -0.5232841372489929,
|
| 487 |
+
"rewards/margins": 0.059422146528959274,
|
| 488 |
+
"rewards/rejected": -0.5827063322067261,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.9996020692399522,
|
| 493 |
+
"step": 314,
|
| 494 |
+
"total_flos": 0.0,
|
| 495 |
+
"train_loss": 0.678918091354856,
|
| 496 |
+
"train_runtime": 7383.0158,
|
| 497 |
+
"train_samples_per_second": 21.78,
|
| 498 |
+
"train_steps_per_second": 0.043
|
| 499 |
+
}
|
| 500 |
+
],
|
| 501 |
+
"logging_steps": 10,
|
| 502 |
+
"max_steps": 314,
|
| 503 |
+
"num_input_tokens_seen": 0,
|
| 504 |
+
"num_train_epochs": 1,
|
| 505 |
+
"save_steps": 100,
|
| 506 |
+
"stateful_callbacks": {
|
| 507 |
+
"TrainerControl": {
|
| 508 |
+
"args": {
|
| 509 |
+
"should_epoch_stop": false,
|
| 510 |
+
"should_evaluate": false,
|
| 511 |
+
"should_log": false,
|
| 512 |
+
"should_save": true,
|
| 513 |
+
"should_training_stop": true
|
| 514 |
+
},
|
| 515 |
+
"attributes": {}
|
| 516 |
+
}
|
| 517 |
+
},
|
| 518 |
+
"total_flos": 0.0,
|
| 519 |
+
"train_batch_size": 16,
|
| 520 |
+
"trial_name": null,
|
| 521 |
+
"trial_params": null
|
| 522 |
+
}
|