Training in progress, epoch 26, checkpoint

checkpoint-5062/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-5062/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7537cfebe86df135864c9df96748f010a80e7d04b68b6709f9abe2433a3b0e6d
+size 143269386

checkpoint-5062/added_tokens.json

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

checkpoint-5062/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9717850f3297d448d3f00d3dfc7c8fdb16d6a14c0b9cd9773489ca02ca293b86
-size 224537202
+oid sha256:010b2daac01a0aa719f9d15076af8eb691425227d681ca08cf3b49f0388a71be
+size 286585234

checkpoint-5062/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:35e3b74cb4233a56871656cceb5aa387e0ed774bd02c051283f8784d385f7f46
+oid sha256:102281996517bddfb6559f5d1acd8d1cd453b94c339f014d30fc82dd96550952
 size 15024

checkpoint-5062/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7eeebe4f3eb32b3ffdfe64e347cb3d2936708ac76a2b9f2f0010d844eac5777b
+oid sha256:dcc1150884832776312cee82282aa2521832a26b94ed42e20d623a313082df07
 size 15024

checkpoint-5062/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0fd530c382c357730a5d95abee0ff4a74014d0b8cd04c21c656de86f97f8a430
+oid sha256:e59e2882d3b8744434a0b33fdbe5b66ae7f491c982c94e2a7018893fa4e4c480
 size 15024

checkpoint-5062/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ebde068d34f95c3336701459f9492cf001922f38844aad42dacd3f5504d4813f
+oid sha256:8d7880a0b6f3b93b70d63b9fc7058cce27e925c204caaa86612f83f65c7f1b04
 size 15024

checkpoint-5062/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-5062/tokenizer.model

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

checkpoint-5062/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-5062/trainer_state.json

@@ -9,780 +9,716 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 0.7527768015861511,
+      "epoch": 0.53,
       "learning_rate": 0.0005,
-      "loss": 1.7217,
+      "loss": 1.43,
       "step": 100
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_accuracy": 0.5738974358974359,
-      "eval_loss": 1.6209890842437744,
-      "eval_runtime": 5.1723,
-      "eval_samples_per_second": 96.668,
-      "eval_steps_per_second": 12.18,
+      "epoch": 1.0,
+      "eval_accuracy": 0.6161538461538462,
+      "eval_loss": 1.2682808637619019,
+      "eval_runtime": 4.85,
+      "eval_samples_per_second": 103.093,
+      "eval_steps_per_second": 12.99,
       "step": 187
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_exact_match": 18.2,
-      "eval_f1": 27.000476190476196,
+      "epoch": 1.0,
+      "eval_exact_match": 26.6,
+      "eval_f1": 38.8657142857143,
       "step": 187
     },
     {
-      "epoch": 1.0666666666666667,
-      "grad_norm": 0.7723727226257324,
+      "epoch": 1.07,
       "learning_rate": 0.0005,
-      "loss": 1.615,
+      "loss": 1.2637,
       "step": 200
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.8481244444847107,
       "learning_rate": 0.0005,
-      "loss": 1.2605,
+      "loss": 1.0285,
       "step": 300
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5706666666666667,
-      "eval_loss": 1.6909353733062744,
-      "eval_runtime": 5.4194,
-      "eval_samples_per_second": 92.26,
-      "eval_steps_per_second": 11.625,
+      "eval_accuracy": 0.6129230769230769,
+      "eval_loss": 1.3220478296279907,
+      "eval_runtime": 4.4227,
+      "eval_samples_per_second": 113.052,
+      "eval_steps_per_second": 14.245,
       "step": 375
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 25.9057142857143,
+      "eval_exact_match": 24.2,
+      "eval_f1": 37.12603174603175,
       "step": 375
     },
     {
-      "epoch": 2.1333333333333333,
-      "grad_norm": 0.810580313205719,
+      "epoch": 2.13,
       "learning_rate": 0.0005,
-      "loss": 1.1991,
+      "loss": 0.9478,
       "step": 400
     },
     {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 0.8971414566040039,
+      "epoch": 2.67,
       "learning_rate": 0.0005,
-      "loss": 0.9141,
+      "loss": 0.7318,
       "step": 500
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_accuracy": 0.5657435897435897,
-      "eval_loss": 1.831854224205017,
-      "eval_runtime": 4.9933,
-      "eval_samples_per_second": 100.134,
-      "eval_steps_per_second": 12.617,
+      "epoch": 3.0,
+      "eval_accuracy": 0.6075897435897436,
+      "eval_loss": 1.4644758701324463,
+      "eval_runtime": 4.8316,
+      "eval_samples_per_second": 103.486,
+      "eval_steps_per_second": 13.039,
       "step": 562
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_exact_match": 16.0,
-      "eval_f1": 23.940000000000012,
+      "epoch": 3.0,
+      "eval_exact_match": 22.8,
+      "eval_f1": 36.51031746031747,
       "step": 562
     },
     {
       "epoch": 3.2,
-      "grad_norm": 0.7627168297767639,
       "learning_rate": 0.0005,
-      "loss": 0.8634,
+      "loss": 0.6922,
       "step": 600
     },
     {
-      "epoch": 3.7333333333333334,
-      "grad_norm": 0.7384397387504578,
+      "epoch": 3.73,
       "learning_rate": 0.0005,
-      "loss": 0.7223,
+      "loss": 0.5898,
       "step": 700
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5637948717948718,
-      "eval_loss": 1.9418420791625977,
-      "eval_runtime": 5.3476,
-      "eval_samples_per_second": 93.499,
-      "eval_steps_per_second": 11.781,
+      "eval_accuracy": 0.604974358974359,
+      "eval_loss": 1.54537832736969,
+      "eval_runtime": 4.119,
+      "eval_samples_per_second": 121.388,
+      "eval_steps_per_second": 15.295,
       "step": 750
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 16.6,
-      "eval_f1": 24.14476190476191,
+      "eval_exact_match": 15.4,
+      "eval_f1": 25.034444444444457,
       "step": 750
     },
     {
-      "epoch": 4.266666666666667,
-      "grad_norm": 0.7065069079399109,
+      "epoch": 4.27,
       "learning_rate": 0.0005,
-      "loss": 0.6786,
+      "loss": 0.5599,
       "step": 800
     },
     {
       "epoch": 4.8,
-      "grad_norm": 0.7625908851623535,
       "learning_rate": 0.0005,
-      "loss": 0.6434,
+      "loss": 0.5309,
       "step": 900
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_accuracy": 0.5602051282051282,
-      "eval_loss": 2.093710422515869,
-      "eval_runtime": 5.1796,
-      "eval_samples_per_second": 96.532,
-      "eval_steps_per_second": 12.163,
+      "epoch": 5.0,
+      "eval_accuracy": 0.6025641025641025,
+      "eval_loss": 1.6439025402069092,
+      "eval_runtime": 4.3353,
+      "eval_samples_per_second": 115.332,
+      "eval_steps_per_second": 14.532,
       "step": 937
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_exact_match": 17.2,
-      "eval_f1": 24.822857142857156,
+      "epoch": 5.0,
+      "eval_exact_match": 21.6,
+      "eval_f1": 34.31112554112555,
       "step": 937
     },
     {
-      "epoch": 5.333333333333333,
-      "grad_norm": 0.6257635951042175,
+      "epoch": 5.33,
       "learning_rate": 0.0005,
-      "loss": 0.6022,
+      "loss": 0.5011,
       "step": 1000
     },
     {
-      "epoch": 5.866666666666667,
-      "grad_norm": 0.7330732345581055,
+      "epoch": 5.87,
       "learning_rate": 0.0005,
-      "loss": 0.5957,
+      "loss": 0.4985,
       "step": 1100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5573333333333333,
-      "eval_loss": 2.17874813079834,
-      "eval_runtime": 5.3048,
-      "eval_samples_per_second": 94.255,
-      "eval_steps_per_second": 11.876,
+      "eval_accuracy": 0.6034358974358974,
+      "eval_loss": 1.722017526626587,
+      "eval_runtime": 4.2182,
+      "eval_samples_per_second": 118.533,
+      "eval_steps_per_second": 14.935,
       "step": 1125
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 15.2,
-      "eval_f1": 22.341269841269852,
+      "eval_exact_match": 22.4,
+      "eval_f1": 34.31126984126985,
       "step": 1125
     },
     {
       "epoch": 6.4,
-      "grad_norm": 0.5348079204559326,
       "learning_rate": 0.0005,
-      "loss": 0.5583,
+      "loss": 0.4686,
       "step": 1200
     },
     {
-      "epoch": 6.933333333333334,
-      "grad_norm": 0.4842449426651001,
+      "epoch": 6.93,
       "learning_rate": 0.0005,
-      "loss": 0.5728,
+      "loss": 0.5091,
       "step": 1300
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_accuracy": 0.558,
-      "eval_loss": 2.2602250576019287,
-      "eval_runtime": 5.068,
-      "eval_samples_per_second": 98.658,
-      "eval_steps_per_second": 12.431,
+      "epoch": 7.0,
+      "eval_accuracy": 0.6008205128205129,
+      "eval_loss": 1.800759196281433,
+      "eval_runtime": 4.8504,
+      "eval_samples_per_second": 103.083,
+      "eval_steps_per_second": 12.989,
       "step": 1312
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_exact_match": 15.8,
-      "eval_f1": 22.874285714285733,
+      "epoch": 7.0,
+      "eval_exact_match": 0.2,
+      "eval_f1": 2.1497326649958226,
       "step": 1312
     },
     {
-      "epoch": 7.466666666666667,
-      "grad_norm": 0.5476555824279785,
+      "epoch": 7.47,
       "learning_rate": 0.0005,
-      "loss": 0.5339,
+      "loss": 0.4576,
       "step": 1400
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.8545846939086914,
       "learning_rate": 0.0005,
-      "loss": 0.5596,
+      "loss": 0.4796,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5559487179487179,
-      "eval_loss": 2.2222671508789062,
-      "eval_runtime": 5.0471,
-      "eval_samples_per_second": 99.067,
-      "eval_steps_per_second": 12.482,
+      "eval_accuracy": 0.6001025641025641,
+      "eval_loss": 1.7781891822814941,
+      "eval_runtime": 4.4147,
+      "eval_samples_per_second": 113.257,
+      "eval_steps_per_second": 14.27,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 15.2,
-      "eval_f1": 23.320000000000007,
+      "eval_exact_match": 19.2,
+      "eval_f1": 29.972380952380963,
       "step": 1500
     },
     {
-      "epoch": 8.533333333333333,
-      "grad_norm": 0.5809034705162048,
+      "epoch": 8.53,
       "learning_rate": 0.0005,
-      "loss": 0.5218,
+      "loss": 0.4453,
       "step": 1600
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_accuracy": 0.5538461538461539,
-      "eval_loss": 2.2796590328216553,
-      "eval_runtime": 5.0384,
-      "eval_samples_per_second": 99.238,
-      "eval_steps_per_second": 12.504,
+      "epoch": 9.0,
+      "eval_accuracy": 0.5984615384615385,
+      "eval_loss": 1.8255126476287842,
+      "eval_runtime": 4.1057,
+      "eval_samples_per_second": 121.781,
+      "eval_steps_per_second": 15.344,
       "step": 1687
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_exact_match": 13.2,
-      "eval_f1": 20.368109668109675,
+      "epoch": 9.0,
+      "eval_exact_match": 17.8,
+      "eval_f1": 29.10082251082252,
       "step": 1687
     },
     {
-      "epoch": 9.066666666666666,
-      "grad_norm": 0.48757097125053406,
+      "epoch": 9.07,
       "learning_rate": 0.0005,
-      "loss": 0.5444,
+      "loss": 0.4701,
       "step": 1700
     },
     {
       "epoch": 9.6,
-      "grad_norm": 0.5225056409835815,
       "learning_rate": 0.0005,
-      "loss": 0.5254,
+      "loss": 0.448,
       "step": 1800
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5561025641025641,
-      "eval_loss": 2.275782823562622,
-      "eval_runtime": 5.158,
-      "eval_samples_per_second": 96.937,
-      "eval_steps_per_second": 12.214,
+      "eval_accuracy": 0.5931282051282051,
+      "eval_loss": 1.7979023456573486,
+      "eval_runtime": 4.4695,
+      "eval_samples_per_second": 111.87,
+      "eval_steps_per_second": 14.096,
       "step": 1875
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 14.0,
-      "eval_f1": 21.65190476190477,
+      "eval_exact_match": 18.8,
+      "eval_f1": 28.28873015873017,
       "step": 1875
     },
     {
-      "epoch": 10.133333333333333,
-      "grad_norm": 0.7123950719833374,
+      "epoch": 10.13,
       "learning_rate": 0.0005,
-      "loss": 0.5304,
+      "loss": 0.4641,
       "step": 1900
     },
     {
-      "epoch": 10.666666666666666,
-      "grad_norm": 0.48232755064964294,
+      "epoch": 10.67,
       "learning_rate": 0.0005,
-      "loss": 0.526,
+      "loss": 0.4522,
       "step": 2000
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_accuracy": 0.5544102564102564,
-      "eval_loss": 2.2686431407928467,
-      "eval_runtime": 5.0517,
-      "eval_samples_per_second": 98.976,
-      "eval_steps_per_second": 12.471,
+      "epoch": 11.0,
+      "eval_accuracy": 0.595948717948718,
+      "eval_loss": 1.8272360563278198,
+      "eval_runtime": 4.1203,
+      "eval_samples_per_second": 121.35,
+      "eval_steps_per_second": 15.29,
       "step": 2062
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_exact_match": 13.0,
-      "eval_f1": 20.387460317460324,
+      "epoch": 11.0,
+      "eval_exact_match": 17.2,
+      "eval_f1": 28.030476190476207,
       "step": 2062
     },
     {
       "epoch": 11.2,
-      "grad_norm": 0.6355456113815308,
       "learning_rate": 0.0005,
-      "loss": 0.5335,
+      "loss": 0.4555,
       "step": 2100
     },
     {
-      "epoch": 11.733333333333333,
-      "grad_norm": 0.7310680150985718,
+      "epoch": 11.73,
       "learning_rate": 0.0005,
-      "loss": 0.5396,
+      "loss": 0.4552,
       "step": 2200
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.5533846153846154,
-      "eval_loss": 2.2792105674743652,
-      "eval_runtime": 5.1276,
-      "eval_samples_per_second": 97.511,
-      "eval_steps_per_second": 12.286,
+      "eval_accuracy": 0.5946153846153847,
+      "eval_loss": 1.8670085668563843,
+      "eval_runtime": 4.2129,
+      "eval_samples_per_second": 118.684,
+      "eval_steps_per_second": 14.954,
       "step": 2250
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 13.4,
-      "eval_f1": 20.533650793650807,
+      "eval_exact_match": 17.8,
+      "eval_f1": 28.45285714285715,
       "step": 2250
     },
     {
-      "epoch": 12.266666666666667,
-      "grad_norm": 0.8953878879547119,
+      "epoch": 12.27,
       "learning_rate": 0.0005,
-      "loss": 0.5383,
+      "loss": 0.4492,
       "step": 2300
     },
     {
       "epoch": 12.8,
-      "grad_norm": 0.8268033266067505,
       "learning_rate": 0.0005,
-      "loss": 0.5466,
+      "loss": 0.4551,
       "step": 2400
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_accuracy": 0.5500512820512821,
-      "eval_loss": 2.2915709018707275,
-      "eval_runtime": 5.0459,
-      "eval_samples_per_second": 99.089,
-      "eval_steps_per_second": 12.485,
+      "epoch": 13.0,
+      "eval_accuracy": 0.595025641025641,
+      "eval_loss": 1.8706276416778564,
+      "eval_runtime": 4.2102,
+      "eval_samples_per_second": 118.759,
+      "eval_steps_per_second": 14.964,
       "step": 2437
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_exact_match": 14.0,
-      "eval_f1": 21.203650793650798,
+      "epoch": 13.0,
+      "eval_exact_match": 19.0,
+      "eval_f1": 28.634603174603175,
       "step": 2437
     },
     {
-      "epoch": 13.333333333333334,
-      "grad_norm": 0.9471064209938049,
+      "epoch": 13.33,
       "learning_rate": 0.0005,
-      "loss": 0.5417,
+      "loss": 0.4505,
       "step": 2500
     },
     {
-      "epoch": 13.866666666666667,
-      "grad_norm": 1.0178123712539673,
+      "epoch": 13.87,
       "learning_rate": 0.0005,
-      "loss": 0.556,
+      "loss": 0.4559,
       "step": 2600
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5496923076923077,
-      "eval_loss": 2.3209521770477295,
-      "eval_runtime": 5.4611,
-      "eval_samples_per_second": 91.557,
-      "eval_steps_per_second": 11.536,
+      "eval_accuracy": 0.5925128205128205,
+      "eval_loss": 1.8730883598327637,
+      "eval_runtime": 4.1209,
+      "eval_samples_per_second": 121.333,
+      "eval_steps_per_second": 15.288,
       "step": 2625
     },
     {
       "epoch": 14.0,
-      "eval_exact_match": 12.6,
-      "eval_f1": 20.174790764790767,
+      "eval_exact_match": 16.4,
+      "eval_f1": 25.989682539682548,
       "step": 2625
     },
     {
       "epoch": 14.4,
-      "grad_norm": 0.750542402267456,
       "learning_rate": 0.0005,
-      "loss": 0.5322,
+      "loss": 0.4455,
       "step": 2700
     },
     {
-      "epoch": 14.933333333333334,
-      "grad_norm": 0.8345568776130676,
+      "epoch": 14.93,
       "learning_rate": 0.0005,
-      "loss": 0.5501,
+      "loss": 0.4581,
       "step": 2800
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_accuracy": 0.5464615384615384,
-      "eval_loss": 2.3918418884277344,
-      "eval_runtime": 5.1243,
-      "eval_samples_per_second": 97.574,
-      "eval_steps_per_second": 12.294,
+      "epoch": 15.0,
+      "eval_accuracy": 0.5932307692307692,
+      "eval_loss": 1.8531454801559448,
+      "eval_runtime": 4.4124,
+      "eval_samples_per_second": 113.316,
+      "eval_steps_per_second": 14.278,
       "step": 2812
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_exact_match": 12.4,
-      "eval_f1": 19.851428571428578,
+      "epoch": 15.0,
+      "eval_exact_match": 17.8,
+      "eval_f1": 28.33587301587303,
       "step": 2812
     },
     {
-      "epoch": 15.466666666666667,
-      "grad_norm": 0.60535728931427,
+      "epoch": 15.47,
       "learning_rate": 0.0005,
-      "loss": 0.5174,
+      "loss": 0.4358,
       "step": 2900
     },
     {
       "epoch": 16.0,
-      "grad_norm": 0.7003186941146851,
       "learning_rate": 0.0005,
-      "loss": 0.5461,
+      "loss": 0.4535,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5475897435897435,
-      "eval_loss": 2.3702445030212402,
-      "eval_runtime": 5.6883,
-      "eval_samples_per_second": 87.9,
-      "eval_steps_per_second": 11.075,
+      "eval_accuracy": 0.5923076923076923,
+      "eval_loss": 1.949161410331726,
+      "eval_runtime": 4.8341,
+      "eval_samples_per_second": 103.433,
+      "eval_steps_per_second": 13.033,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_exact_match": 12.6,
-      "eval_f1": 19.352380952380955,
+      "eval_exact_match": 17.4,
+      "eval_f1": 27.222698412698417,
       "step": 3000
     },
     {
-      "epoch": 16.533333333333335,
-      "grad_norm": 0.9538315534591675,
+      "epoch": 16.53,
       "learning_rate": 0.0005,
-      "loss": 0.5032,
+      "loss": 0.4308,
       "step": 3100
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_accuracy": 0.5461025641025641,
-      "eval_loss": 2.443984270095825,
-      "eval_runtime": 5.1587,
-      "eval_samples_per_second": 96.923,
-      "eval_steps_per_second": 12.212,
+      "epoch": 17.0,
+      "eval_accuracy": 0.5915384615384616,
+      "eval_loss": 1.8943943977355957,
+      "eval_runtime": 4.2634,
+      "eval_samples_per_second": 117.279,
+      "eval_steps_per_second": 14.777,
       "step": 3187
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_exact_match": 11.6,
-      "eval_f1": 19.255238095238102,
+      "epoch": 17.0,
+      "eval_exact_match": 17.4,
+      "eval_f1": 26.819047619047623,
       "step": 3187
     },
     {
-      "epoch": 17.066666666666666,
-      "grad_norm": 0.7299863696098328,
+      "epoch": 17.07,
       "learning_rate": 0.0005,
-      "loss": 0.5267,
+      "loss": 0.4467,
       "step": 3200
     },
     {
       "epoch": 17.6,
-      "grad_norm": 0.49411433935165405,
       "learning_rate": 0.0005,
-      "loss": 0.501,
+      "loss": 0.4312,
       "step": 3300
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5468205128205128,
-      "eval_loss": 2.4240849018096924,
-      "eval_runtime": 5.0346,
-      "eval_samples_per_second": 99.313,
-      "eval_steps_per_second": 12.513,
+      "eval_accuracy": 0.5903589743589743,
+      "eval_loss": 1.9315078258514404,
+      "eval_runtime": 4.84,
+      "eval_samples_per_second": 103.306,
+      "eval_steps_per_second": 13.017,
       "step": 3375
     },
     {
       "epoch": 18.0,
-      "eval_exact_match": 12.8,
-      "eval_f1": 19.46714285714286,
+      "eval_exact_match": 16.6,
+      "eval_f1": 26.37841269841271,
       "step": 3375
     },
     {
-      "epoch": 18.133333333333333,
-      "grad_norm": 0.5169008374214172,
+      "epoch": 18.13,
       "learning_rate": 0.0005,
-      "loss": 0.5077,
+      "loss": 0.4418,
       "step": 3400
     },
     {
-      "epoch": 18.666666666666668,
-      "grad_norm": 0.4064069092273712,
+      "epoch": 18.67,
       "learning_rate": 0.0005,
-      "loss": 0.4975,
+      "loss": 0.4372,
       "step": 3500
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_accuracy": 0.5450769230769231,
-      "eval_loss": 2.4591097831726074,
-      "eval_runtime": 5.7434,
-      "eval_samples_per_second": 87.056,
-      "eval_steps_per_second": 10.969,
+      "epoch": 19.0,
+      "eval_accuracy": 0.589948717948718,
+      "eval_loss": 1.9201208353042603,
+      "eval_runtime": 4.1791,
+      "eval_samples_per_second": 119.642,
+      "eval_steps_per_second": 15.075,
       "step": 3562
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_exact_match": 12.8,
-      "eval_f1": 19.07666666666667,
+      "epoch": 19.0,
+      "eval_exact_match": 17.0,
+      "eval_f1": 26.170000000000005,
       "step": 3562
     },
     {
       "epoch": 19.2,
-      "grad_norm": 0.362119197845459,
       "learning_rate": 0.0005,
-      "loss": 0.5028,
+      "loss": 0.4385,
       "step": 3600
     },
     {
-      "epoch": 19.733333333333334,
-      "grad_norm": 0.45963579416275024,
+      "epoch": 19.73,
       "learning_rate": 0.0005,
-      "loss": 0.4977,
+      "loss": 0.4359,
       "step": 3700
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.545025641025641,
-      "eval_loss": 2.4378092288970947,
-      "eval_runtime": 5.0227,
-      "eval_samples_per_second": 99.548,
-      "eval_steps_per_second": 12.543,
+      "eval_accuracy": 0.5895384615384616,
+      "eval_loss": 1.9752758741378784,
+      "eval_runtime": 4.6876,
+      "eval_samples_per_second": 106.664,
+      "eval_steps_per_second": 13.44,
       "step": 3750
     },
     {
       "epoch": 20.0,
-      "eval_exact_match": 12.6,
-      "eval_f1": 19.655873015873016,
+      "eval_exact_match": 16.6,
+      "eval_f1": 26.214126984126988,
       "step": 3750
     },
     {
-      "epoch": 20.266666666666666,
-      "grad_norm": 0.6107614636421204,
+      "epoch": 20.27,
       "learning_rate": 0.0005,
-      "loss": 0.4964,
+      "loss": 0.4294,
       "step": 3800
     },
     {
       "epoch": 20.8,
-      "grad_norm": 0.5394221544265747,
       "learning_rate": 0.0005,
-      "loss": 0.501,
+      "loss": 0.4363,
       "step": 3900
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_accuracy": 0.5438461538461539,
-      "eval_loss": 2.448828935623169,
-      "eval_runtime": 5.1176,
-      "eval_samples_per_second": 97.702,
-      "eval_steps_per_second": 12.31,
+      "epoch": 21.0,
+      "eval_accuracy": 0.5876923076923077,
+      "eval_loss": 1.993224024772644,
+      "eval_runtime": 4.8398,
+      "eval_samples_per_second": 103.31,
+      "eval_steps_per_second": 13.017,
       "step": 3937
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_exact_match": 12.6,
-      "eval_f1": 18.367460317460317,
+      "epoch": 21.0,
+      "eval_exact_match": 15.6,
+      "eval_f1": 24.723492063492078,
       "step": 3937
     },
     {
-      "epoch": 21.333333333333332,
-      "grad_norm": 0.35812634229660034,
+      "epoch": 21.33,
       "learning_rate": 0.0005,
-      "loss": 0.4964,
+      "loss": 0.4291,
       "step": 4000
     },
     {
-      "epoch": 21.866666666666667,
-      "grad_norm": 0.4076957404613495,
+      "epoch": 21.87,
       "learning_rate": 0.0005,
-      "loss": 0.5133,
+      "loss": 0.4404,
       "step": 4100
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.5428717948717948,
-      "eval_loss": 2.548297882080078,
-      "eval_runtime": 6.0053,
-      "eval_samples_per_second": 83.26,
-      "eval_steps_per_second": 10.491,
+      "eval_accuracy": 0.5865641025641025,
+      "eval_loss": 2.0326027870178223,
+      "eval_runtime": 4.2213,
+      "eval_samples_per_second": 118.447,
+      "eval_steps_per_second": 14.924,
       "step": 4125
     },
     {
       "epoch": 22.0,
-      "eval_exact_match": 11.6,
-      "eval_f1": 17.79587301587302,
+      "eval_exact_match": 15.2,
+      "eval_f1": 24.99412698412699,
       "step": 4125
     },
     {
       "epoch": 22.4,
-      "grad_norm": 0.5148049592971802,
       "learning_rate": 0.0005,
-      "loss": 0.4957,
+      "loss": 0.4303,
       "step": 4200
     },
     {
-      "epoch": 22.933333333333334,
-      "grad_norm": 0.7316868305206299,
+      "epoch": 22.93,
       "learning_rate": 0.0005,
-      "loss": 0.5154,
+      "loss": 0.4436,
       "step": 4300
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_accuracy": 0.5442051282051282,
-      "eval_loss": 2.460996389389038,
-      "eval_runtime": 5.0409,
-      "eval_samples_per_second": 99.189,
-      "eval_steps_per_second": 12.498,
+      "epoch": 23.0,
+      "eval_accuracy": 0.5847692307692308,
+      "eval_loss": 2.000843048095703,
+      "eval_runtime": 4.4158,
+      "eval_samples_per_second": 113.229,
+      "eval_steps_per_second": 14.267,
       "step": 4312
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_exact_match": 13.4,
-      "eval_f1": 19.424285714285713,
+      "epoch": 23.0,
+      "eval_exact_match": 14.4,
+      "eval_f1": 23.444285714285712,
       "step": 4312
     },
     {
-      "epoch": 23.466666666666665,
-      "grad_norm": 0.7490749955177307,
+      "epoch": 23.47,
       "learning_rate": 0.0005,
-      "loss": 0.5006,
+      "loss": 0.4268,
       "step": 4400
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.7351626753807068,
       "learning_rate": 0.0005,
-      "loss": 0.5257,
+      "loss": 0.4438,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.5448717948717948,
-      "eval_loss": 2.503119945526123,
-      "eval_runtime": 4.9846,
-      "eval_samples_per_second": 100.308,
-      "eval_steps_per_second": 12.639,
+      "eval_accuracy": 0.5877435897435898,
+      "eval_loss": 2.018592596054077,
+      "eval_runtime": 4.2112,
+      "eval_samples_per_second": 118.731,
+      "eval_steps_per_second": 14.96,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_exact_match": 12.2,
-      "eval_f1": 18.472222222222225,
+      "eval_exact_match": 16.0,
+      "eval_f1": 24.778412698412712,
       "step": 4500
     },
     {
-      "epoch": 24.533333333333335,
-      "grad_norm": 0.5879587531089783,
+      "epoch": 24.53,
       "learning_rate": 0.0005,
-      "loss": 0.5065,
+      "loss": 0.4233,
       "step": 4600
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_accuracy": 0.5415384615384615,
-      "eval_loss": 2.510695695877075,
-      "eval_runtime": 5.7813,
-      "eval_samples_per_second": 86.485,
-      "eval_steps_per_second": 10.897,
+      "epoch": 25.0,
+      "eval_accuracy": 0.5862564102564103,
+      "eval_loss": 2.04520320892334,
+      "eval_runtime": 4.1216,
+      "eval_samples_per_second": 121.312,
+      "eval_steps_per_second": 15.285,
       "step": 4687
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_exact_match": 13.2,
-      "eval_f1": 19.150952380952386,
+      "epoch": 25.0,
+      "eval_exact_match": 14.6,
+      "eval_f1": 22.220476190476194,
       "step": 4687
     },
     {
-      "epoch": 25.066666666666666,
-      "grad_norm": 0.61104816198349,
+      "epoch": 25.07,
       "learning_rate": 0.0005,
-      "loss": 0.5273,
+      "loss": 0.4388,
       "step": 4700
     },
     {
       "epoch": 25.6,
-      "grad_norm": 1.318751335144043,
       "learning_rate": 0.0005,
-      "loss": 0.5177,
+      "loss": 0.4237,
       "step": 4800
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.5402051282051282,
-      "eval_loss": 2.5010507106781006,
-      "eval_runtime": 5.3385,
-      "eval_samples_per_second": 93.66,
-      "eval_steps_per_second": 11.801,
+      "eval_accuracy": 0.5842564102564103,
+      "eval_loss": 2.0520122051239014,
+      "eval_runtime": 4.2989,
+      "eval_samples_per_second": 116.31,
+      "eval_steps_per_second": 14.655,
       "step": 4875
     },
     {
       "epoch": 26.0,
-      "eval_exact_match": 12.0,
-      "eval_f1": 18.683665223665226,
+      "eval_exact_match": 15.8,
+      "eval_f1": 24.92222222222223,
       "step": 4875
     },
     {
-      "epoch": 26.133333333333333,
-      "grad_norm": 0.8964791893959045,
+      "epoch": 26.13,
       "learning_rate": 0.0005,
-      "loss": 0.5267,
+      "loss": 0.4362,
       "step": 4900
     },
     {
-      "epoch": 26.666666666666668,
-      "grad_norm": 0.6350087523460388,
+      "epoch": 26.67,
       "learning_rate": 0.0005,
-      "loss": 0.5115,
+      "loss": 0.4289,
       "step": 5000
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_accuracy": 0.5407179487179488,
-      "eval_loss": 2.547609567642212,
-      "eval_runtime": 5.1686,
-      "eval_samples_per_second": 96.738,
-      "eval_steps_per_second": 12.189,
+      "epoch": 27.0,
+      "eval_accuracy": 0.5827692307692308,
+      "eval_loss": 2.0817410945892334,
+      "eval_runtime": 4.4121,
+      "eval_samples_per_second": 113.325,
+      "eval_steps_per_second": 14.279,
       "step": 5062
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_exact_match": 12.4,
-      "eval_f1": 18.38857142857143,
+      "epoch": 27.0,
+      "eval_exact_match": 13.6,
+      "eval_f1": 21.646507936507945,
       "step": 5062
     }
   ],
   "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": 8.816747905810432e+16,
-  "train_batch_size": 1,
+  "total_flos": 1.8889352134262784e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-5062/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2776615b5656a131d464a15a3d5283d9bdf82687bc8ffc8d243c778d1158123d
-size 5304
+oid sha256:aeaafcc180d545c4c7bc83b3ef040967ab10c6394ea7a4c1dc01fb190e8fc740
+size 4728