Training in progress, epoch 12, checkpoint

checkpoint-2437/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-2437/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:306441147de88c91c631a674be3cd21b0dd0c70fe4f024c2ce6c658000289954
+size 143269386

checkpoint-2437/added_tokens.json

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

checkpoint-2437/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:214dcd5978f5c7376b2cfaabaf1044a6877c901ed2b0e3a48bfa8a4e28ba8f4d
-size 224537202
+oid sha256:5198ddef8174c7977b90db3db1de6d9cc58ec6a39fa58e98e0f5aea09fb0b19c
+size 286585234

checkpoint-2437/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2544149d522ceedff75ddc3f79827209bf49f81af684fa808fc93861de953a46
+oid sha256:1f62cfbecee25d844fc8be936984e69921f6551aff1848f2848d43ec56928b29
 size 15024

checkpoint-2437/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55c6a994f89f83634e7f8e6a78d219a44fcedd8960208ed50dadfdc80f9e17e3
+oid sha256:52323c0b55c2e122711ce5c06f254b5a8bbfcf626506bcc9558f4c42d7af10d5
 size 15024

checkpoint-2437/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1df24c86f5a32f24930bec27d59824adaf3f7ef1dc7c5d62bba448c0738868a
+oid sha256:a3e0bad4b81eaa806addcccff7106e177dd9345293214ba37c5bc9663991b314
 size 15024

checkpoint-2437/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2affd23a0bbde6427b435745972a69cb9e8a6039f9c862c9c14089b4e94fa089
+oid sha256:5da30da00ffa7679d30f4389b18e4e3303b191ea3bd2541446817f21a1cda067
 size 15024

checkpoint-2437/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-2437/tokenizer.model

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

checkpoint-2437/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-2437/trainer_state.json

@@ -9,388 +9,350 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 0.64143306016922,
+      "epoch": 0.53,
       "learning_rate": 0.0001,
-      "loss": 1.7779,
+      "loss": 1.598,
       "step": 100
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_accuracy": 0.5734871794871795,
-      "eval_loss": 1.6191272735595703,
-      "eval_runtime": 5.393,
-      "eval_samples_per_second": 92.713,
-      "eval_steps_per_second": 11.682,
+      "epoch": 1.0,
+      "eval_accuracy": 0.6146666666666667,
+      "eval_loss": 1.2691738605499268,
+      "eval_runtime": 4.3939,
+      "eval_samples_per_second": 113.793,
+      "eval_steps_per_second": 14.338,
       "step": 187
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_exact_match": 18.0,
-      "eval_f1": 28.280476190476193,
+      "epoch": 1.0,
+      "eval_exact_match": 24.8,
+      "eval_f1": 38.10857142857143,
       "step": 187
     },
     {
-      "epoch": 1.0666666666666667,
-      "grad_norm": 0.7118939161300659,
+      "epoch": 1.07,
       "learning_rate": 0.0001,
-      "loss": 1.6349,
+      "loss": 1.2894,
       "step": 200
     },
     {
       "epoch": 1.6,
-      "grad_norm": 1.0694799423217773,
       "learning_rate": 0.0001,
-      "loss": 1.4887,
+      "loss": 1.1923,
       "step": 300
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5735897435897436,
-      "eval_loss": 1.6172032356262207,
-      "eval_runtime": 5.0425,
-      "eval_samples_per_second": 99.158,
-      "eval_steps_per_second": 12.494,
+      "eval_accuracy": 0.6175897435897436,
+      "eval_loss": 1.2732893228530884,
+      "eval_runtime": 4.5482,
+      "eval_samples_per_second": 109.934,
+      "eval_steps_per_second": 13.852,
       "step": 375
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 19.8,
-      "eval_f1": 28.77095238095238,
+      "eval_exact_match": 25.6,
+      "eval_f1": 39.04746031746033,
       "step": 375
     },
     {
-      "epoch": 2.1333333333333333,
-      "grad_norm": 1.28153395652771,
+      "epoch": 2.13,
       "learning_rate": 0.0001,
-      "loss": 1.4304,
+      "loss": 1.1174,
       "step": 400
     },
     {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 1.3190444707870483,
+      "epoch": 2.67,
       "learning_rate": 0.0001,
-      "loss": 1.2346,
+      "loss": 0.9732,
       "step": 500
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_accuracy": 0.5706153846153846,
-      "eval_loss": 1.6950472593307495,
-      "eval_runtime": 5.1981,
-      "eval_samples_per_second": 96.188,
-      "eval_steps_per_second": 12.12,
+      "epoch": 3.0,
+      "eval_accuracy": 0.6135897435897436,
+      "eval_loss": 1.33958101272583,
+      "eval_runtime": 4.0988,
+      "eval_samples_per_second": 121.986,
+      "eval_steps_per_second": 15.37,
       "step": 562
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_exact_match": 19.4,
-      "eval_f1": 28.241904761904756,
+      "epoch": 3.0,
+      "eval_exact_match": 25.4,
+      "eval_f1": 37.65698412698413,
       "step": 562
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.5040416717529297,
       "learning_rate": 0.0001,
-      "loss": 1.1583,
+      "loss": 0.9111,
       "step": 600
     },
     {
-      "epoch": 3.7333333333333334,
-      "grad_norm": 1.512290358543396,
+      "epoch": 3.73,
       "learning_rate": 0.0001,
-      "loss": 1.0104,
+      "loss": 0.7763,
       "step": 700
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5655897435897436,
-      "eval_loss": 1.7944799661636353,
-      "eval_runtime": 5.3441,
-      "eval_samples_per_second": 93.561,
-      "eval_steps_per_second": 11.789,
+      "eval_accuracy": 0.6104102564102564,
+      "eval_loss": 1.4358161687850952,
+      "eval_runtime": 4.4028,
+      "eval_samples_per_second": 113.565,
+      "eval_steps_per_second": 14.309,
       "step": 750
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.639523809523823,
+      "eval_exact_match": 25.2,
+      "eval_f1": 37.8809523809524,
       "step": 750
     },
     {
-      "epoch": 4.266666666666667,
-      "grad_norm": 1.698282241821289,
+      "epoch": 4.27,
       "learning_rate": 0.0001,
-      "loss": 0.9275,
+      "loss": 0.7118,
       "step": 800
     },
     {
       "epoch": 4.8,
-      "grad_norm": 1.6007401943206787,
       "learning_rate": 0.0001,
-      "loss": 0.843,
+      "loss": 0.6498,
       "step": 900
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_accuracy": 0.5632820512820513,
-      "eval_loss": 1.9219905138015747,
-      "eval_runtime": 5.2659,
-      "eval_samples_per_second": 94.951,
-      "eval_steps_per_second": 11.964,
+      "epoch": 5.0,
+      "eval_accuracy": 0.6051794871794872,
+      "eval_loss": 1.5630384683609009,
+      "eval_runtime": 4.1079,
+      "eval_samples_per_second": 121.717,
+      "eval_steps_per_second": 15.336,
       "step": 937
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_exact_match": 15.8,
-      "eval_f1": 24.631904761904778,
+      "epoch": 5.0,
+      "eval_exact_match": 23.4,
+      "eval_f1": 35.55111111111113,
       "step": 937
     },
     {
-      "epoch": 5.333333333333333,
-      "grad_norm": 1.5375592708587646,
+      "epoch": 5.33,
       "learning_rate": 0.0001,
-      "loss": 0.7516,
+      "loss": 0.5911,
       "step": 1000
     },
     {
-      "epoch": 5.866666666666667,
-      "grad_norm": 1.718989372253418,
+      "epoch": 5.87,
       "learning_rate": 0.0001,
-      "loss": 0.7199,
+      "loss": 0.57,
       "step": 1100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5616923076923077,
-      "eval_loss": 2.016322612762451,
-      "eval_runtime": 5.116,
-      "eval_samples_per_second": 97.734,
-      "eval_steps_per_second": 12.314,
+      "eval_accuracy": 0.6031282051282051,
+      "eval_loss": 1.6599360704421997,
+      "eval_runtime": 4.2005,
+      "eval_samples_per_second": 119.033,
+      "eval_steps_per_second": 14.998,
       "step": 1125
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.39285714285715,
+      "eval_exact_match": 22.6,
+      "eval_f1": 34.44650793650795,
       "step": 1125
     },
     {
       "epoch": 6.4,
-      "grad_norm": 1.8337936401367188,
       "learning_rate": 0.0001,
-      "loss": 0.6378,
+      "loss": 0.517,
       "step": 1200
     },
     {
-      "epoch": 6.933333333333334,
-      "grad_norm": 1.9019217491149902,
+      "epoch": 6.93,
       "learning_rate": 0.0001,
-      "loss": 0.6432,
+      "loss": 0.5253,
       "step": 1300
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_accuracy": 0.5616410256410257,
-      "eval_loss": 2.1283841133117676,
-      "eval_runtime": 5.1352,
-      "eval_samples_per_second": 97.367,
-      "eval_steps_per_second": 12.268,
+      "epoch": 7.0,
+      "eval_accuracy": 0.6027179487179487,
+      "eval_loss": 1.7479698657989502,
+      "eval_runtime": 4.4276,
+      "eval_samples_per_second": 112.929,
+      "eval_steps_per_second": 14.229,
       "step": 1312
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 22.042857142857148,
+      "epoch": 7.0,
+      "eval_exact_match": 22.2,
+      "eval_f1": 33.87303030303032,
       "step": 1312
     },
     {
-      "epoch": 7.466666666666667,
-      "grad_norm": 1.246907114982605,
+      "epoch": 7.47,
       "learning_rate": 0.0001,
-      "loss": 0.578,
+      "loss": 0.4784,
       "step": 1400
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.6256760358810425,
       "learning_rate": 0.0001,
-      "loss": 0.596,
+      "loss": 0.4958,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5603589743589743,
-      "eval_loss": 2.170966386795044,
-      "eval_runtime": 5.0636,
-      "eval_samples_per_second": 98.744,
-      "eval_steps_per_second": 12.442,
+      "eval_accuracy": 0.602051282051282,
+      "eval_loss": 1.8059930801391602,
+      "eval_runtime": 4.1898,
+      "eval_samples_per_second": 119.337,
+      "eval_steps_per_second": 15.036,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.71636363636364,
+      "eval_exact_match": 21.8,
+      "eval_f1": 32.95365079365081,
       "step": 1500
     },
     {
-      "epoch": 8.533333333333333,
-      "grad_norm": 1.3106721639633179,
+      "epoch": 8.53,
       "learning_rate": 0.0001,
-      "loss": 0.543,
+      "loss": 0.4521,
       "step": 1600
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_accuracy": 0.5595384615384615,
-      "eval_loss": 2.262348175048828,
-      "eval_runtime": 5.3198,
-      "eval_samples_per_second": 93.989,
-      "eval_steps_per_second": 11.843,
+      "epoch": 9.0,
+      "eval_accuracy": 0.6013333333333334,
+      "eval_loss": 1.8598920106887817,
+      "eval_runtime": 4.9021,
+      "eval_samples_per_second": 101.996,
+      "eval_steps_per_second": 12.852,
       "step": 1687
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.34507936507938,
+      "epoch": 9.0,
+      "eval_exact_match": 22.6,
+      "eval_f1": 33.89858585858587,
       "step": 1687
     },
     {
-      "epoch": 9.066666666666666,
-      "grad_norm": 1.0499460697174072,
+      "epoch": 9.07,
       "learning_rate": 0.0001,
-      "loss": 0.5644,
+      "loss": 0.4694,
       "step": 1700
     },
     {
       "epoch": 9.6,
-      "grad_norm": 0.9217801690101624,
       "learning_rate": 0.0001,
-      "loss": 0.5314,
+      "loss": 0.443,
       "step": 1800
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5583589743589744,
-      "eval_loss": 2.268125057220459,
-      "eval_runtime": 5.457,
-      "eval_samples_per_second": 91.626,
-      "eval_steps_per_second": 11.545,
+      "eval_accuracy": 0.6013333333333334,
+      "eval_loss": 1.9468085765838623,
+      "eval_runtime": 4.4033,
+      "eval_samples_per_second": 113.552,
+      "eval_steps_per_second": 14.308,
       "step": 1875
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 23.46095238095238,
+      "eval_exact_match": 21.8,
+      "eval_f1": 32.7704761904762,
       "step": 1875
     },
     {
-      "epoch": 10.133333333333333,
-      "grad_norm": 0.969529926776886,
+      "epoch": 10.13,
       "learning_rate": 0.0001,
-      "loss": 0.534,
+      "loss": 0.4513,
       "step": 1900
     },
     {
-      "epoch": 10.666666666666666,
-      "grad_norm": 0.9540187120437622,
+      "epoch": 10.67,
       "learning_rate": 0.0001,
-      "loss": 0.5171,
+      "loss": 0.439,
       "step": 2000
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_accuracy": 0.5584615384615385,
-      "eval_loss": 2.3312978744506836,
-      "eval_runtime": 5.2187,
-      "eval_samples_per_second": 95.81,
-      "eval_steps_per_second": 12.072,
+      "epoch": 11.0,
+      "eval_accuracy": 0.6014871794871794,
+      "eval_loss": 1.9500343799591064,
+      "eval_runtime": 4.0917,
+      "eval_samples_per_second": 122.197,
+      "eval_steps_per_second": 15.397,
       "step": 2062
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_exact_match": 16.2,
-      "eval_f1": 23.968888888888895,
+      "epoch": 11.0,
+      "eval_exact_match": 22.4,
+      "eval_f1": 33.80857142857144,
       "step": 2062
     },
     {
       "epoch": 11.2,
-      "grad_norm": 0.8946124911308289,
       "learning_rate": 0.0001,
-      "loss": 0.5208,
+      "loss": 0.4372,
       "step": 2100
     },
     {
-      "epoch": 11.733333333333333,
-      "grad_norm": 0.8362706899642944,
+      "epoch": 11.73,
       "learning_rate": 0.0001,
-      "loss": 0.512,
+      "loss": 0.433,
       "step": 2200
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.559025641025641,
-      "eval_loss": 2.364483594894409,
-      "eval_runtime": 5.0957,
-      "eval_samples_per_second": 98.122,
-      "eval_steps_per_second": 12.363,
+      "eval_accuracy": 0.6021025641025641,
+      "eval_loss": 1.9103901386260986,
+      "eval_runtime": 4.4069,
+      "eval_samples_per_second": 113.459,
+      "eval_steps_per_second": 14.296,
       "step": 2250
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.015714285714285,
+      "eval_exact_match": 21.4,
+      "eval_f1": 33.23952380952382,
       "step": 2250
     },
     {
-      "epoch": 12.266666666666667,
-      "grad_norm": 0.7387551665306091,
+      "epoch": 12.27,
       "learning_rate": 0.0001,
-      "loss": 0.5038,
+      "loss": 0.4301,
       "step": 2300
     },
     {
       "epoch": 12.8,
-      "grad_norm": 0.8488346338272095,
       "learning_rate": 0.0001,
-      "loss": 0.503,
+      "loss": 0.4323,
       "step": 2400
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_accuracy": 0.5588205128205128,
-      "eval_loss": 2.3642191886901855,
-      "eval_runtime": 5.4807,
-      "eval_samples_per_second": 91.23,
-      "eval_steps_per_second": 11.495,
+      "epoch": 13.0,
+      "eval_accuracy": 0.6001025641025641,
+      "eval_loss": 2.0078930854797363,
+      "eval_runtime": 4.1244,
+      "eval_samples_per_second": 121.229,
+      "eval_steps_per_second": 15.275,
       "step": 2437
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 23.92238095238096,
+      "epoch": 13.0,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.54033189033191,
       "step": 2437
     }
   ],
   "logging_steps": 100,
   "max_steps": 9350,
-  "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": 4.464687839392563e+16,
-  "train_batch_size": 2,
+  "total_flos": 9.094174110803558e+16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-2437/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d667d5e0e40fe951f9fe0d2cd7fb4682a608b09af78a113a245307211adedfbc
-size 5304
+oid sha256:dfdad39e2291c20f000713e6b820dbeb747acc6873c9eec22bd69c854755aa38
+size 4728