Training in progress, epoch 8, checkpoint

checkpoint-2732/adapter_config.json

@@ -1,6 +1,6 @@
 {
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen1.5-4B",
+  "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,

checkpoint-2732/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3636d7310d2eee08339b3df2dd51b7f1e3735ff62f0c46174768cd4096eaba66
+size 143269386

checkpoint-2732/added_tokens.json

@@ -1,5 +1,5 @@
 {
-  "<|endoftext|>": 151643,
-  "<|im_end|>": 151645,
-  "<|im_start|>": 151644
+  "</s>": 2,
+  "<s>": 1,
+  "<unk>": 0
 }

checkpoint-2732/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f91c66a28ff4b14692200c1cdf78d5105eb03cc935834ffb998c78700969bbed
-size 224537202
+oid sha256:43d7c93f433dc4eab28c1d1c3f790c094ba6c3dfbc72cb8de4aa448563b85573
+size 286585234

checkpoint-2732/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:99a32ae35fea79138331484ff5e3c1adc6f999086d773a33382998884c1d872a
+oid sha256:73172b09e8c12d2acf2f84acc161101b67cec2631659bd3dfa9da39e896574cd
 size 15024

checkpoint-2732/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:88743c3671796c3ec4431180dc4b60fb519ed4a342a971e36c50b4dac1442203
+oid sha256:69302127f77d17f83db641fa8c127ed0c573de0fa949eaba576e6d783f1af669
 size 15024

checkpoint-2732/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa3e80271ca0c3464cb676ad7a3d3bd51b10f2b6d7e823456edb934d4695f85f
+oid sha256:42d946093effa862f64adc2040bca8949640567d12d139bc19688a1aa2399305
 size 15024

checkpoint-2732/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cedb3af708b24697d951cb3f62a16d25e3228fbe9abe0caa3962bd93fdd2b29d
+oid sha256:b96ba6d56474359329036d444886c0cd540ef25e7f14c01cfe0944a2ff316caf
 size 15024

checkpoint-2732/special_tokens_map.json

@@ -1,14 +1,6 @@
 {
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "eos_token": {
-    "content": "<|endoftext|>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "pad_token": "<|endoftext|>"
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "</s>",
+  "unk_token": "<unk>"
 }

checkpoint-2732/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

checkpoint-2732/tokenizer_config.json

@@ -1,24 +1,23 @@
 {
-  "add_prefix_space": false,
   "added_tokens_decoder": {
-    "151643": {
-      "content": "<|endoftext|>",
+    "0": {
+      "content": "<unk>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151644": {
-      "content": "<|im_start|>",
+    "1": {
+      "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151645": {
-      "content": "<|im_end|>",
+    "2": {
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -26,18 +25,16 @@
       "special": true
     }
   },
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "bos_token": null,
-  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
   "clean_up_tokenization_spaces": false,
-  "eos_token": "<|endoftext|>",
-  "errors": "replace",
-  "model_max_length": 32768,
-  "pad_token": "<|endoftext|>",
-  "split_special_tokens": false,
-  "tokenizer_class": "Qwen2Tokenizer",
-  "unk_token": null
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "left",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true
 }

checkpoint-2732/trainer_state.json

@@ -9,342 +9,301 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.29282576866764276,
-      "grad_norm": 0.26115021109580994,
+      "epoch": 0.29,
       "learning_rate": 3e-05,
-      "loss": 2.0719,
+      "loss": 1.574,
       "step": 100
     },
     {
-      "epoch": 0.5856515373352855,
-      "grad_norm": 0.2679530680179596,
+      "epoch": 0.59,
       "learning_rate": 3e-05,
-      "loss": 1.969,
+      "loss": 1.474,
       "step": 200
     },
     {
-      "epoch": 0.8784773060029283,
-      "grad_norm": 0.2916224002838135,
+      "epoch": 0.88,
       "learning_rate": 3e-05,
-      "loss": 1.9686,
+      "loss": 1.4104,
       "step": 300
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_accuracy": 0.43743589743589745,
-      "eval_loss": 3.6918885707855225,
-      "eval_runtime": 5.2974,
-      "eval_samples_per_second": 94.385,
-      "eval_steps_per_second": 11.893,
+      "epoch": 1.0,
+      "eval_accuracy": 0.4536923076923077,
+      "eval_loss": 3.3574578762054443,
+      "eval_runtime": 4.4299,
+      "eval_samples_per_second": 112.869,
+      "eval_steps_per_second": 14.221,
       "step": 341
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_bleu": 0.1527908371131025,
-      "eval_exact_match": 0.012,
-      "eval_prefix_exact_match": 0.026,
+      "epoch": 1.0,
+      "eval_bleu": 0.08840557310176213,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.058,
       "step": 341
     },
     {
-      "epoch": 1.171303074670571,
-      "grad_norm": 0.3534347712993622,
+      "epoch": 1.17,
       "learning_rate": 3e-05,
-      "loss": 1.9603,
+      "loss": 1.3889,
       "step": 400
     },
     {
-      "epoch": 1.4641288433382138,
-      "grad_norm": 0.3338424563407898,
+      "epoch": 1.46,
       "learning_rate": 3e-05,
-      "loss": 1.9235,
+      "loss": 1.387,
       "step": 500
     },
     {
-      "epoch": 1.7569546120058566,
-      "grad_norm": 0.4033390283584595,
+      "epoch": 1.76,
       "learning_rate": 3e-05,
-      "loss": 1.9337,
+      "loss": 1.389,
       "step": 600
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.44764102564102565,
-      "eval_loss": 3.7484843730926514,
-      "eval_runtime": 5.0649,
-      "eval_samples_per_second": 98.718,
-      "eval_steps_per_second": 12.438,
+      "eval_accuracy": 0.4543589743589744,
+      "eval_loss": 3.4179794788360596,
+      "eval_runtime": 4.1997,
+      "eval_samples_per_second": 119.055,
+      "eval_steps_per_second": 15.001,
       "step": 683
     },
     {
       "epoch": 2.0,
-      "eval_bleu": 0.21311912714932216,
-      "eval_exact_match": 0.02,
-      "eval_prefix_exact_match": 0.032,
+      "eval_bleu": 0.09820569565497561,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 683
     },
     {
-      "epoch": 2.049780380673499,
-      "grad_norm": 0.38882532715797424,
+      "epoch": 2.05,
       "learning_rate": 3e-05,
-      "loss": 1.9267,
+      "loss": 1.3725,
       "step": 700
     },
     {
-      "epoch": 2.342606149341142,
-      "grad_norm": 0.44310709834098816,
+      "epoch": 2.34,
       "learning_rate": 3e-05,
-      "loss": 1.8823,
+      "loss": 1.3588,
       "step": 800
     },
     {
-      "epoch": 2.6354319180087846,
-      "grad_norm": 0.5196680426597595,
+      "epoch": 2.64,
       "learning_rate": 3e-05,
-      "loss": 1.901,
+      "loss": 1.348,
       "step": 900
     },
     {
-      "epoch": 2.9282576866764276,
-      "grad_norm": 0.5293437242507935,
+      "epoch": 2.93,
       "learning_rate": 3e-05,
-      "loss": 1.9033,
+      "loss": 1.3414,
       "step": 1000
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_accuracy": 0.4495897435897436,
-      "eval_loss": 3.882551670074463,
-      "eval_runtime": 5.1314,
-      "eval_samples_per_second": 97.439,
-      "eval_steps_per_second": 12.277,
+      "epoch": 3.0,
+      "eval_accuracy": 0.45476923076923076,
+      "eval_loss": 3.511923313140869,
+      "eval_runtime": 4.1985,
+      "eval_samples_per_second": 119.089,
+      "eval_steps_per_second": 15.005,
       "step": 1024
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_bleu": 0.24307488023362148,
-      "eval_exact_match": 0.016,
-      "eval_prefix_exact_match": 0.032,
+      "epoch": 3.0,
+      "eval_bleu": 0.09889831624589397,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 1024
     },
     {
-      "epoch": 3.22108345534407,
-      "grad_norm": 0.552943229675293,
+      "epoch": 3.22,
       "learning_rate": 3e-05,
-      "loss": 1.8529,
+      "loss": 1.3166,
       "step": 1100
     },
     {
-      "epoch": 3.513909224011713,
-      "grad_norm": 0.6421984434127808,
+      "epoch": 3.51,
       "learning_rate": 3e-05,
-      "loss": 1.8342,
+      "loss": 1.3035,
       "step": 1200
     },
     {
-      "epoch": 3.8067349926793557,
-      "grad_norm": 0.7155198454856873,
+      "epoch": 3.81,
       "learning_rate": 3e-05,
-      "loss": 1.857,
+      "loss": 1.3002,
       "step": 1300
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.4481025641025641,
-      "eval_loss": 3.970107316970825,
-      "eval_runtime": 5.1563,
-      "eval_samples_per_second": 96.969,
-      "eval_steps_per_second": 12.218,
+      "eval_accuracy": 0.45543589743589746,
+      "eval_loss": 3.5288071632385254,
+      "eval_runtime": 4.4066,
+      "eval_samples_per_second": 113.466,
+      "eval_steps_per_second": 14.297,
       "step": 1366
     },
     {
       "epoch": 4.0,
-      "eval_bleu": 0.25309406645138827,
-      "eval_exact_match": 0.016,
-      "eval_prefix_exact_match": 0.044,
+      "eval_bleu": 0.11209345982903633,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.054,
       "step": 1366
     },
     {
-      "epoch": 4.099560761346998,
-      "grad_norm": 0.7837833166122437,
+      "epoch": 4.1,
       "learning_rate": 3e-05,
-      "loss": 1.8284,
+      "loss": 1.2929,
       "step": 1400
     },
     {
-      "epoch": 4.392386530014641,
-      "grad_norm": 0.784609854221344,
+      "epoch": 4.39,
       "learning_rate": 3e-05,
-      "loss": 1.7884,
+      "loss": 1.2693,
       "step": 1500
     },
     {
-      "epoch": 4.685212298682284,
-      "grad_norm": 0.834195613861084,
+      "epoch": 4.69,
       "learning_rate": 3e-05,
-      "loss": 1.7841,
+      "loss": 1.2697,
       "step": 1600
     },
     {
-      "epoch": 4.978038067349927,
-      "grad_norm": 0.8902062177658081,
+      "epoch": 4.98,
       "learning_rate": 3e-05,
-      "loss": 1.8042,
+      "loss": 1.2574,
       "step": 1700
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_accuracy": 0.4472820512820513,
-      "eval_loss": 4.117065906524658,
-      "eval_runtime": 5.2376,
-      "eval_samples_per_second": 95.464,
-      "eval_steps_per_second": 12.028,
+      "epoch": 5.0,
+      "eval_accuracy": 0.45394871794871794,
+      "eval_loss": 3.6893365383148193,
+      "eval_runtime": 4.1122,
+      "eval_samples_per_second": 121.591,
+      "eval_steps_per_second": 15.32,
       "step": 1707
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_bleu": 0.2531049955279983,
-      "eval_exact_match": 0.014,
-      "eval_prefix_exact_match": 0.042,
+      "epoch": 5.0,
+      "eval_bleu": 0.12076940663377221,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 1707
     },
     {
-      "epoch": 5.270863836017569,
-      "grad_norm": 0.9495383501052856,
+      "epoch": 5.27,
       "learning_rate": 3e-05,
-      "loss": 1.7493,
+      "loss": 1.2097,
       "step": 1800
     },
     {
-      "epoch": 5.563689604685212,
-      "grad_norm": 0.9647482633590698,
+      "epoch": 5.56,
       "learning_rate": 3e-05,
-      "loss": 1.7334,
+      "loss": 1.2188,
       "step": 1900
     },
     {
-      "epoch": 5.856515373352855,
-      "grad_norm": 0.9825767874717712,
+      "epoch": 5.86,
       "learning_rate": 3e-05,
-      "loss": 1.7443,
+      "loss": 1.2258,
       "step": 2000
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.446974358974359,
-      "eval_loss": 4.183700084686279,
-      "eval_runtime": 5.2674,
-      "eval_samples_per_second": 94.923,
-      "eval_steps_per_second": 11.96,
+      "eval_accuracy": 0.45615384615384613,
+      "eval_loss": 3.725893497467041,
+      "eval_runtime": 4.0989,
+      "eval_samples_per_second": 121.984,
+      "eval_steps_per_second": 15.37,
       "step": 2049
     },
     {
       "epoch": 6.0,
-      "eval_bleu": 0.23587359955369594,
-      "eval_exact_match": 0.012,
-      "eval_prefix_exact_match": 0.04,
+      "eval_bleu": 0.11884272898137646,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.058,
       "step": 2049
     },
     {
-      "epoch": 6.149341142020498,
-      "grad_norm": 1.0430296659469604,
+      "epoch": 6.15,
       "learning_rate": 3e-05,
-      "loss": 1.6959,
+      "loss": 1.205,
       "step": 2100
     },
     {
-      "epoch": 6.44216691068814,
-      "grad_norm": 1.0777732133865356,
+      "epoch": 6.44,
       "learning_rate": 3e-05,
-      "loss": 1.6795,
+      "loss": 1.1643,
       "step": 2200
     },
     {
-      "epoch": 6.734992679355783,
-      "grad_norm": 1.0959677696228027,
+      "epoch": 6.73,
       "learning_rate": 3e-05,
-      "loss": 1.7019,
+      "loss": 1.1844,
       "step": 2300
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_accuracy": 0.4461538461538462,
-      "eval_loss": 4.260408401489258,
-      "eval_runtime": 5.178,
-      "eval_samples_per_second": 96.563,
-      "eval_steps_per_second": 12.167,
+      "epoch": 7.0,
+      "eval_accuracy": 0.45594871794871794,
+      "eval_loss": 3.724449396133423,
+      "eval_runtime": 4.8158,
+      "eval_samples_per_second": 103.825,
+      "eval_steps_per_second": 13.082,
       "step": 2390
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_bleu": 0.2008158466316808,
-      "eval_exact_match": 0.006,
-      "eval_prefix_exact_match": 0.042,
+      "epoch": 7.0,
+      "eval_bleu": 0.12407896336995003,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.048,
       "step": 2390
     },
     {
-      "epoch": 7.027818448023426,
-      "grad_norm": 1.0649549961090088,
+      "epoch": 7.03,
       "learning_rate": 3e-05,
-      "loss": 1.6812,
+      "loss": 1.1771,
       "step": 2400
     },
     {
-      "epoch": 7.320644216691068,
-      "grad_norm": 1.294122576713562,
+      "epoch": 7.32,
       "learning_rate": 3e-05,
-      "loss": 1.6226,
+      "loss": 1.1218,
       "step": 2500
     },
     {
-      "epoch": 7.613469985358711,
-      "grad_norm": 1.2657020092010498,
+      "epoch": 7.61,
       "learning_rate": 3e-05,
-      "loss": 1.6404,
+      "loss": 1.1491,
       "step": 2600
     },
     {
-      "epoch": 7.906295754026354,
-      "grad_norm": 1.1963133811950684,
+      "epoch": 7.91,
       "learning_rate": 3e-05,
-      "loss": 1.6305,
+      "loss": 1.1363,
       "step": 2700
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.44148717948717947,
-      "eval_loss": 4.406544208526611,
-      "eval_runtime": 5.3355,
-      "eval_samples_per_second": 93.711,
-      "eval_steps_per_second": 11.808,
+      "eval_accuracy": 0.4543589743589744,
+      "eval_loss": 3.8138701915740967,
+      "eval_runtime": 4.1963,
+      "eval_samples_per_second": 119.152,
+      "eval_steps_per_second": 15.013,
       "step": 2732
     },
     {
       "epoch": 8.0,
-      "eval_bleu": 0.24435500637461924,
-      "eval_exact_match": 0.008,
-      "eval_prefix_exact_match": 0.038,
+      "eval_bleu": 0.11947629327387163,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.046,
       "step": 2732
     }
   ],
   "logging_steps": 100,
   "max_steps": 17050,
-  "num_input_tokens_seen": 0,
   "num_train_epochs": 50,
   "save_steps": 500,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 2.3459846877190554e+17,
-  "train_batch_size": 1,
+  "total_flos": 5.410663755069522e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-2732/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9cef08c7f10bf396c269dfc26104fd5a94756f5431d6448312805ef8e298380
-size 5304
+oid sha256:28a33ee2cb1c54e667d59907eed3cc20ab64c3794314f21d17db553ff2af83d6
+size 4728