Training in progress, epoch 28, checkpoint

checkpoint-5250/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-5250/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e18e0660abd49fa16615254a6ab2fcd2bed8af2589d8f25b8c21f4db26113354
+size 143269386

checkpoint-5250/added_tokens.json

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

checkpoint-5250/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fab06f728ff2aecdad64765852ace0b96377df390de8fad96f81787762c96f44
-size 224537202
+oid sha256:479940bf0b9c7d45253eab74ad149addaa56418a5aa92439180137ae100af532
+size 286585234

checkpoint-5250/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:73b317296299b37630b8ae23f9a6931f429fcbc64e0a1e3cc68284905d123441
+oid sha256:5a90fd9aab58259d92c6b1067d138c2ffbe3d562e0ad81ae4e45ed2d2df434b5
 size 15024

checkpoint-5250/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7fd786cd89f7ea328384ff7c7acf84c38979bdb770b744c893917185048c0562
+oid sha256:20296eb00673d1db0dd98f35b343de884546bcf9c1f9085655c33bdc9c5a37cf
 size 15024

checkpoint-5250/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:020749a24a043ec764552f090e5a0b24653fd2f71080d900a1d036cce8670806
+oid sha256:a93daa012f81a8c45414a599132a113dc68cbe03ea824a5038d2a5b92f31a14e
 size 15024

checkpoint-5250/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3c5005e6849e1d684466ea07c2991968e83deee7bb870b59bd383c732329c24b
+oid sha256:b0016fa3f6cb7b5f234d0113590228e370278a9a482338ee8544b26ebd12da88
 size 15024

checkpoint-5250/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-5250/tokenizer.model

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

checkpoint-5250/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-5250/trainer_state.json

@@ -9,809 +9,743 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5333333333333333,
-      "grad_norm": 0.5194531679153442,
+      "epoch": 0.53,
       "learning_rate": 0.0003,
-      "loss": 1.7259,
+      "loss": 1.4433,
       "step": 100
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_accuracy": 0.5753333333333334,
-      "eval_loss": 1.6105104684829712,
-      "eval_runtime": 5.418,
-      "eval_samples_per_second": 92.285,
-      "eval_steps_per_second": 11.628,
+      "epoch": 1.0,
+      "eval_accuracy": 0.6165641025641025,
+      "eval_loss": 1.2625622749328613,
+      "eval_runtime": 4.3923,
+      "eval_samples_per_second": 113.835,
+      "eval_steps_per_second": 14.343,
       "step": 187
     },
     {
-      "epoch": 0.9973333333333333,
-      "eval_exact_match": 17.2,
-      "eval_f1": 26.890173160173163,
+      "epoch": 1.0,
+      "eval_exact_match": 26.4,
+      "eval_f1": 39.52571428571429,
       "step": 187
     },
     {
-      "epoch": 1.0666666666666667,
-      "grad_norm": 0.6734262704849243,
+      "epoch": 1.07,
       "learning_rate": 0.0003,
-      "loss": 1.6156,
+      "loss": 1.2659,
       "step": 200
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.7422465682029724,
       "learning_rate": 0.0003,
-      "loss": 1.3358,
+      "loss": 1.0761,
       "step": 300
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5722564102564103,
-      "eval_loss": 1.6464648246765137,
-      "eval_runtime": 5.1856,
-      "eval_samples_per_second": 96.422,
-      "eval_steps_per_second": 12.149,
+      "eval_accuracy": 0.615897435897436,
+      "eval_loss": 1.295995831489563,
+      "eval_runtime": 4.4145,
+      "eval_samples_per_second": 113.264,
+      "eval_steps_per_second": 14.271,
       "step": 375
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 19.0,
-      "eval_f1": 27.762380952380955,
+      "eval_exact_match": 25.4,
+      "eval_f1": 38.36603174603176,
       "step": 375
     },
     {
-      "epoch": 2.1333333333333333,
-      "grad_norm": 0.828364908695221,
+      "epoch": 2.13,
       "learning_rate": 0.0003,
-      "loss": 1.2607,
+      "loss": 0.9947,
       "step": 400
     },
     {
-      "epoch": 2.6666666666666665,
-      "grad_norm": 0.9025559425354004,
+      "epoch": 2.67,
       "learning_rate": 0.0003,
-      "loss": 0.9854,
+      "loss": 0.7909,
       "step": 500
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_accuracy": 0.5707692307692308,
-      "eval_loss": 1.7620763778686523,
-      "eval_runtime": 5.2589,
-      "eval_samples_per_second": 95.077,
-      "eval_steps_per_second": 11.98,
+      "epoch": 3.0,
+      "eval_accuracy": 0.6093846153846154,
+      "eval_loss": 1.42075777053833,
+      "eval_runtime": 4.4832,
+      "eval_samples_per_second": 111.529,
+      "eval_steps_per_second": 14.053,
       "step": 562
     },
     {
-      "epoch": 2.997333333333333,
-      "eval_exact_match": 18.2,
-      "eval_f1": 27.53142857142857,
+      "epoch": 3.0,
+      "eval_exact_match": 23.6,
+      "eval_f1": 36.76761904761906,
       "step": 562
     },
     {
       "epoch": 3.2,
-      "grad_norm": 0.8217537999153137,
       "learning_rate": 0.0003,
-      "loss": 0.9099,
+      "loss": 0.7343,
       "step": 600
     },
     {
-      "epoch": 3.7333333333333334,
-      "grad_norm": 0.8315110206604004,
+      "epoch": 3.73,
       "learning_rate": 0.0003,
-      "loss": 0.7508,
+      "loss": 0.6117,
       "step": 700
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5663076923076923,
-      "eval_loss": 1.9380683898925781,
-      "eval_runtime": 5.3338,
-      "eval_samples_per_second": 93.742,
-      "eval_steps_per_second": 11.811,
+      "eval_accuracy": 0.6065128205128205,
+      "eval_loss": 1.509896159172058,
+      "eval_runtime": 4.8185,
+      "eval_samples_per_second": 103.766,
+      "eval_steps_per_second": 13.075,
       "step": 750
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.611904761904768,
+      "eval_exact_match": 22.6,
+      "eval_f1": 35.788253968253976,
       "step": 750
     },
     {
-      "epoch": 4.266666666666667,
-      "grad_norm": 0.6763857007026672,
+      "epoch": 4.27,
       "learning_rate": 0.0003,
-      "loss": 0.6912,
+      "loss": 0.5688,
       "step": 800
     },
     {
       "epoch": 4.8,
-      "grad_norm": 0.8408166766166687,
       "learning_rate": 0.0003,
-      "loss": 0.6396,
+      "loss": 0.535,
       "step": 900
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_accuracy": 0.5634358974358974,
-      "eval_loss": 1.9925932884216309,
-      "eval_runtime": 5.3717,
-      "eval_samples_per_second": 93.08,
-      "eval_steps_per_second": 11.728,
+      "epoch": 5.0,
+      "eval_accuracy": 0.6073333333333333,
+      "eval_loss": 1.6383951902389526,
+      "eval_runtime": 4.1044,
+      "eval_samples_per_second": 121.82,
+      "eval_steps_per_second": 15.349,
       "step": 937
     },
     {
-      "epoch": 4.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.313809523809528,
+      "epoch": 5.0,
+      "eval_exact_match": 23.8,
+      "eval_f1": 37.04809523809523,
       "step": 937
     },
     {
-      "epoch": 5.333333333333333,
-      "grad_norm": 0.6721271276473999,
+      "epoch": 5.33,
       "learning_rate": 0.0003,
-      "loss": 0.5959,
+      "loss": 0.4997,
       "step": 1000
     },
     {
-      "epoch": 5.866666666666667,
-      "grad_norm": 0.9258321523666382,
+      "epoch": 5.87,
       "learning_rate": 0.0003,
-      "loss": 0.5833,
+      "loss": 0.4917,
       "step": 1100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5611794871794872,
-      "eval_loss": 2.101524829864502,
-      "eval_runtime": 5.4208,
-      "eval_samples_per_second": 92.238,
-      "eval_steps_per_second": 11.622,
+      "eval_accuracy": 0.6064102564102564,
+      "eval_loss": 1.6917529106140137,
+      "eval_runtime": 4.8299,
+      "eval_samples_per_second": 103.522,
+      "eval_steps_per_second": 13.044,
       "step": 1125
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 25.444285714285726,
+      "eval_exact_match": 24.2,
+      "eval_f1": 36.030317460317455,
       "step": 1125
     },
     {
       "epoch": 6.4,
-      "grad_norm": 0.6029295921325684,
       "learning_rate": 0.0003,
-      "loss": 0.545,
+      "loss": 0.4605,
       "step": 1200
     },
     {
-      "epoch": 6.933333333333334,
-      "grad_norm": 0.5719736814498901,
+      "epoch": 6.93,
       "learning_rate": 0.0003,
-      "loss": 0.5567,
+      "loss": 0.4746,
       "step": 1300
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_accuracy": 0.5607179487179487,
-      "eval_loss": 2.1645195484161377,
-      "eval_runtime": 5.1974,
-      "eval_samples_per_second": 96.202,
-      "eval_steps_per_second": 12.121,
+      "epoch": 7.0,
+      "eval_accuracy": 0.6048205128205129,
+      "eval_loss": 1.7547425031661987,
+      "eval_runtime": 4.2021,
+      "eval_samples_per_second": 118.988,
+      "eval_steps_per_second": 14.993,
       "step": 1312
     },
     {
-      "epoch": 6.997333333333334,
-      "eval_exact_match": 15.0,
-      "eval_f1": 23.618095238095247,
+      "epoch": 7.0,
+      "eval_exact_match": 23.8,
+      "eval_f1": 36.49507936507937,
       "step": 1312
     },
     {
-      "epoch": 7.466666666666667,
-      "grad_norm": 0.49924588203430176,
+      "epoch": 7.47,
       "learning_rate": 0.0003,
-      "loss": 0.5205,
+      "loss": 0.4432,
       "step": 1400
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.5360875129699707,
       "learning_rate": 0.0003,
-      "loss": 0.5411,
+      "loss": 0.463,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5614358974358974,
-      "eval_loss": 2.2039756774902344,
-      "eval_runtime": 5.0936,
-      "eval_samples_per_second": 98.162,
-      "eval_steps_per_second": 12.368,
+      "eval_accuracy": 0.6056923076923076,
+      "eval_loss": 1.7707523107528687,
+      "eval_runtime": 4.821,
+      "eval_samples_per_second": 103.713,
+      "eval_steps_per_second": 13.068,
       "step": 1500
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 16.8,
-      "eval_f1": 25.132380952380966,
+      "eval_exact_match": 22.4,
+      "eval_f1": 35.834603174603174,
       "step": 1500
     },
     {
-      "epoch": 8.533333333333333,
-      "grad_norm": 0.44560280442237854,
+      "epoch": 8.53,
       "learning_rate": 0.0003,
-      "loss": 0.5028,
+      "loss": 0.4307,
       "step": 1600
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_accuracy": 0.5608205128205128,
-      "eval_loss": 2.236473321914673,
-      "eval_runtime": 5.2122,
-      "eval_samples_per_second": 95.93,
-      "eval_steps_per_second": 12.087,
+      "epoch": 9.0,
+      "eval_accuracy": 0.6051282051282051,
+      "eval_loss": 1.7772654294967651,
+      "eval_runtime": 4.4303,
+      "eval_samples_per_second": 112.86,
+      "eval_steps_per_second": 14.22,
       "step": 1687
     },
     {
-      "epoch": 8.997333333333334,
-      "eval_exact_match": 16.0,
-      "eval_f1": 24.712380952380958,
+      "epoch": 9.0,
+      "eval_exact_match": 24.4,
+      "eval_f1": 36.40761904761904,
       "step": 1687
     },
     {
-      "epoch": 9.066666666666666,
-      "grad_norm": 0.42847225069999695,
+      "epoch": 9.07,
       "learning_rate": 0.0003,
-      "loss": 0.5246,
+      "loss": 0.4694,
       "step": 1700
     },
     {
       "epoch": 9.6,
-      "grad_norm": 0.4186452627182007,
       "learning_rate": 0.0003,
-      "loss": 0.5041,
+      "loss": 0.4323,
       "step": 1800
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5604615384615385,
-      "eval_loss": 2.2862467765808105,
-      "eval_runtime": 5.8141,
-      "eval_samples_per_second": 85.998,
-      "eval_steps_per_second": 10.836,
+      "eval_accuracy": 0.5996923076923077,
+      "eval_loss": 1.7927325963974,
+      "eval_runtime": 4.4251,
+      "eval_samples_per_second": 112.993,
+      "eval_steps_per_second": 14.237,
       "step": 1875
     },
     {
       "epoch": 10.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.683333333333344,
+      "eval_exact_match": 23.2,
+      "eval_f1": 35.76095238095239,
       "step": 1875
     },
     {
-      "epoch": 10.133333333333333,
-      "grad_norm": 0.427721232175827,
+      "epoch": 10.13,
       "learning_rate": 0.0003,
-      "loss": 0.5083,
+      "loss": 0.4392,
       "step": 1900
     },
     {
-      "epoch": 10.666666666666666,
-      "grad_norm": 0.4298701584339142,
+      "epoch": 10.67,
       "learning_rate": 0.0003,
-      "loss": 0.4991,
+      "loss": 0.4301,
       "step": 2000
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_accuracy": 0.5602564102564103,
-      "eval_loss": 2.285149574279785,
-      "eval_runtime": 5.1591,
-      "eval_samples_per_second": 96.916,
-      "eval_steps_per_second": 12.211,
+      "epoch": 11.0,
+      "eval_accuracy": 0.6046153846153847,
+      "eval_loss": 1.820131778717041,
+      "eval_runtime": 4.2028,
+      "eval_samples_per_second": 118.968,
+      "eval_steps_per_second": 14.99,
       "step": 2062
     },
     {
-      "epoch": 10.997333333333334,
-      "eval_exact_match": 14.4,
-      "eval_f1": 22.41619047619048,
+      "epoch": 11.0,
+      "eval_exact_match": 22.6,
+      "eval_f1": 36.275425685425695,
       "step": 2062
     },
     {
       "epoch": 11.2,
-      "grad_norm": 0.40195176005363464,
       "learning_rate": 0.0003,
-      "loss": 0.5053,
+      "loss": 0.4325,
       "step": 2100
     },
     {
-      "epoch": 11.733333333333333,
-      "grad_norm": 0.6030442118644714,
+      "epoch": 11.73,
       "learning_rate": 0.0003,
-      "loss": 0.5048,
+      "loss": 0.4331,
       "step": 2200
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.560974358974359,
-      "eval_loss": 2.2454702854156494,
-      "eval_runtime": 5.1149,
-      "eval_samples_per_second": 97.753,
-      "eval_steps_per_second": 12.317,
+      "eval_accuracy": 0.6050769230769231,
+      "eval_loss": 1.8902928829193115,
+      "eval_runtime": 4.1616,
+      "eval_samples_per_second": 120.145,
+      "eval_steps_per_second": 15.138,
       "step": 2250
     },
     {
       "epoch": 12.0,
-      "eval_exact_match": 16.2,
-      "eval_f1": 24.622857142857157,
+      "eval_exact_match": 23.6,
+      "eval_f1": 35.66952380952382,
       "step": 2250
     },
     {
-      "epoch": 12.266666666666667,
-      "grad_norm": 0.4596252143383026,
+      "epoch": 12.27,
       "learning_rate": 0.0003,
-      "loss": 0.4995,
+      "loss": 0.43,
       "step": 2300
     },
     {
       "epoch": 12.8,
-      "grad_norm": 0.4469389021396637,
       "learning_rate": 0.0003,
-      "loss": 0.5067,
+      "loss": 0.4338,
       "step": 2400
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_accuracy": 0.5591794871794872,
-      "eval_loss": 2.258878707885742,
-      "eval_runtime": 5.1623,
-      "eval_samples_per_second": 96.856,
-      "eval_steps_per_second": 12.204,
+      "epoch": 13.0,
+      "eval_accuracy": 0.6035384615384616,
+      "eval_loss": 1.8489463329315186,
+      "eval_runtime": 4.4116,
+      "eval_samples_per_second": 113.338,
+      "eval_steps_per_second": 14.281,
       "step": 2437
     },
     {
-      "epoch": 12.997333333333334,
-      "eval_exact_match": 14.0,
-      "eval_f1": 22.422539682539693,
+      "epoch": 13.0,
+      "eval_exact_match": 19.6,
+      "eval_f1": 33.558888888888895,
       "step": 2437
     },
     {
-      "epoch": 13.333333333333334,
-      "grad_norm": 0.4875074028968811,
+      "epoch": 13.33,
       "learning_rate": 0.0003,
-      "loss": 0.4998,
+      "loss": 0.4261,
       "step": 2500
     },
     {
-      "epoch": 13.866666666666667,
-      "grad_norm": 0.5969555974006653,
+      "epoch": 13.87,
       "learning_rate": 0.0003,
-      "loss": 0.508,
+      "loss": 0.4331,
       "step": 2600
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.5583589743589744,
-      "eval_loss": 2.2630844116210938,
-      "eval_runtime": 5.2013,
-      "eval_samples_per_second": 96.129,
-      "eval_steps_per_second": 12.112,
+      "eval_accuracy": 0.6024102564102564,
+      "eval_loss": 1.9057968854904175,
+      "eval_runtime": 4.5305,
+      "eval_samples_per_second": 110.363,
+      "eval_steps_per_second": 13.906,
       "step": 2625
     },
     {
       "epoch": 14.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 22.224285714285717,
+      "eval_exact_match": 23.0,
+      "eval_f1": 34.747936507936515,
       "step": 2625
     },
     {
       "epoch": 14.4,
-      "grad_norm": 0.3906678259372711,
       "learning_rate": 0.0003,
-      "loss": 0.5008,
+      "loss": 0.4247,
       "step": 2700
     },
     {
-      "epoch": 14.933333333333334,
-      "grad_norm": 0.6833591461181641,
+      "epoch": 14.93,
       "learning_rate": 0.0003,
-      "loss": 0.514,
+      "loss": 0.4362,
       "step": 2800
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_accuracy": 0.5563589743589744,
-      "eval_loss": 2.277346134185791,
-      "eval_runtime": 5.1497,
-      "eval_samples_per_second": 97.094,
-      "eval_steps_per_second": 12.234,
+      "epoch": 15.0,
+      "eval_accuracy": 0.6004102564102564,
+      "eval_loss": 1.8960095643997192,
+      "eval_runtime": 4.4093,
+      "eval_samples_per_second": 113.396,
+      "eval_steps_per_second": 14.288,
       "step": 2812
     },
     {
-      "epoch": 14.997333333333334,
-      "eval_exact_match": 14.2,
-      "eval_f1": 22.277792207792217,
+      "epoch": 15.0,
+      "eval_exact_match": 12.2,
+      "eval_f1": 20.72142857142858,
       "step": 2812
     },
     {
-      "epoch": 15.466666666666667,
-      "grad_norm": 0.6206196546554565,
+      "epoch": 15.47,
       "learning_rate": 0.0003,
-      "loss": 0.4902,
+      "loss": 0.4188,
       "step": 2900
     },
     {
       "epoch": 16.0,
-      "grad_norm": 0.4821285903453827,
       "learning_rate": 0.0003,
-      "loss": 0.5149,
+      "loss": 0.4371,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5576410256410257,
-      "eval_loss": 2.286123514175415,
-      "eval_runtime": 5.1003,
-      "eval_samples_per_second": 98.034,
-      "eval_steps_per_second": 12.352,
+      "eval_accuracy": 0.6011794871794872,
+      "eval_loss": 1.9080209732055664,
+      "eval_runtime": 4.8301,
+      "eval_samples_per_second": 103.517,
+      "eval_steps_per_second": 13.043,
       "step": 3000
     },
     {
       "epoch": 16.0,
-      "eval_exact_match": 15.8,
-      "eval_f1": 23.748730158730165,
+      "eval_exact_match": 21.6,
+      "eval_f1": 33.10307692307693,
       "step": 3000
     },
     {
-      "epoch": 16.533333333333335,
-      "grad_norm": 0.7102397084236145,
+      "epoch": 16.53,
       "learning_rate": 0.0003,
-      "loss": 0.4835,
+      "loss": 0.4153,
       "step": 3100
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_accuracy": 0.5587692307692308,
-      "eval_loss": 2.2662925720214844,
-      "eval_runtime": 5.8917,
-      "eval_samples_per_second": 84.865,
-      "eval_steps_per_second": 10.693,
+      "epoch": 17.0,
+      "eval_accuracy": 0.601948717948718,
+      "eval_loss": 1.9089771509170532,
+      "eval_runtime": 4.3169,
+      "eval_samples_per_second": 115.823,
+      "eval_steps_per_second": 14.594,
       "step": 3187
     },
     {
-      "epoch": 16.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 22.173809523809535,
+      "epoch": 17.0,
+      "eval_exact_match": 22.8,
+      "eval_f1": 34.82815628815629,
       "step": 3187
     },
     {
-      "epoch": 17.066666666666666,
-      "grad_norm": 0.4603720009326935,
+      "epoch": 17.07,
       "learning_rate": 0.0003,
-      "loss": 0.5031,
+      "loss": 0.4285,
       "step": 3200
     },
     {
       "epoch": 17.6,
-      "grad_norm": 0.45880746841430664,
       "learning_rate": 0.0003,
-      "loss": 0.484,
+      "loss": 0.4151,
       "step": 3300
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5574871794871795,
-      "eval_loss": 2.31445050239563,
-      "eval_runtime": 5.3978,
-      "eval_samples_per_second": 92.63,
-      "eval_steps_per_second": 11.671,
+      "eval_accuracy": 0.6003589743589743,
+      "eval_loss": 1.913118839263916,
+      "eval_runtime": 4.2766,
+      "eval_samples_per_second": 116.915,
+      "eval_steps_per_second": 14.731,
       "step": 3375
     },
     {
       "epoch": 18.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 22.527301587301594,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.15157160620821333,
       "step": 3375
     },
     {
-      "epoch": 18.133333333333333,
-      "grad_norm": 0.4228228032588959,
+      "epoch": 18.13,
       "learning_rate": 0.0003,
-      "loss": 0.4953,
+      "loss": 0.4262,
       "step": 3400
     },
     {
-      "epoch": 18.666666666666668,
-      "grad_norm": 0.42700931429862976,
+      "epoch": 18.67,
       "learning_rate": 0.0003,
-      "loss": 0.4862,
+      "loss": 0.426,
       "step": 3500
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_accuracy": 0.5559487179487179,
-      "eval_loss": 2.2948524951934814,
-      "eval_runtime": 5.1396,
-      "eval_samples_per_second": 97.283,
-      "eval_steps_per_second": 12.258,
+      "epoch": 19.0,
+      "eval_accuracy": 0.574974358974359,
+      "eval_loss": 2.0016062259674072,
+      "eval_runtime": 4.121,
+      "eval_samples_per_second": 121.33,
+      "eval_steps_per_second": 15.288,
       "step": 3562
     },
     {
-      "epoch": 18.997333333333334,
-      "eval_exact_match": 14.0,
-      "eval_f1": 21.949682539682552,
+      "epoch": 19.0,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.0,
       "step": 3562
     },
     {
       "epoch": 19.2,
-      "grad_norm": 0.41856566071510315,
       "learning_rate": 0.0003,
-      "loss": 0.4881,
+      "loss": 2.0542,
       "step": 3600
     },
     {
-      "epoch": 19.733333333333334,
-      "grad_norm": 0.4635602831840515,
+      "epoch": 19.73,
       "learning_rate": 0.0003,
-      "loss": 0.4871,
+      "loss": 0.4475,
       "step": 3700
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.5581025641025641,
-      "eval_loss": 2.3217034339904785,
-      "eval_runtime": 5.423,
-      "eval_samples_per_second": 92.199,
-      "eval_steps_per_second": 11.617,
+      "eval_accuracy": 0.600974358974359,
+      "eval_loss": 1.887317419052124,
+      "eval_runtime": 4.319,
+      "eval_samples_per_second": 115.767,
+      "eval_steps_per_second": 14.587,
       "step": 3750
     },
     {
       "epoch": 20.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.02142857142858,
+      "eval_exact_match": 6.8,
+      "eval_f1": 11.268571428571423,
       "step": 3750
     },
     {
-      "epoch": 20.266666666666666,
-      "grad_norm": 0.3601387143135071,
+      "epoch": 20.27,
       "learning_rate": 0.0003,
-      "loss": 0.4856,
+      "loss": 0.4227,
       "step": 3800
     },
     {
       "epoch": 20.8,
-      "grad_norm": 0.6374465227127075,
       "learning_rate": 0.0003,
-      "loss": 0.4902,
+      "loss": 0.4325,
       "step": 3900
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_accuracy": 0.5571794871794872,
-      "eval_loss": 2.325617790222168,
-      "eval_runtime": 5.1414,
-      "eval_samples_per_second": 97.249,
-      "eval_steps_per_second": 12.253,
+      "epoch": 21.0,
+      "eval_accuracy": 0.6008717948717949,
+      "eval_loss": 1.835098147392273,
+      "eval_runtime": 4.2241,
+      "eval_samples_per_second": 118.368,
+      "eval_steps_per_second": 14.914,
       "step": 3937
     },
     {
-      "epoch": 20.997333333333334,
-      "eval_exact_match": 13.8,
-      "eval_f1": 21.08714285714286,
+      "epoch": 21.0,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.295334700387332,
       "step": 3937
     },
     {
-      "epoch": 21.333333333333332,
-      "grad_norm": 0.7570663690567017,
+      "epoch": 21.33,
       "learning_rate": 0.0003,
-      "loss": 0.4799,
+      "loss": 0.4318,
       "step": 4000
     },
     {
-      "epoch": 21.866666666666667,
-      "grad_norm": 0.8324512839317322,
+      "epoch": 21.87,
       "learning_rate": 0.0003,
-      "loss": 0.492,
+      "loss": 0.4542,
       "step": 4100
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.5558461538461539,
-      "eval_loss": 2.3584253787994385,
-      "eval_runtime": 5.2168,
-      "eval_samples_per_second": 95.845,
-      "eval_steps_per_second": 12.076,
+      "eval_accuracy": 0.598974358974359,
+      "eval_loss": 1.865936040878296,
+      "eval_runtime": 4.4112,
+      "eval_samples_per_second": 113.347,
+      "eval_steps_per_second": 14.282,
       "step": 4125
     },
     {
       "epoch": 22.0,
-      "eval_exact_match": 13.8,
-      "eval_f1": 20.745238095238097,
+      "eval_exact_match": 6.6,
+      "eval_f1": 10.301587301587299,
       "step": 4125
     },
     {
       "epoch": 22.4,
-      "grad_norm": 0.40705105662345886,
       "learning_rate": 0.0003,
-      "loss": 0.4795,
+      "loss": 1.4938,
       "step": 4200
     },
     {
-      "epoch": 22.933333333333334,
-      "grad_norm": 0.38530829548835754,
+      "epoch": 22.93,
       "learning_rate": 0.0003,
-      "loss": 0.4937,
+      "loss": 0.5662,
       "step": 4300
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_accuracy": 0.5557948717948717,
-      "eval_loss": 2.3608150482177734,
-      "eval_runtime": 5.1373,
-      "eval_samples_per_second": 97.326,
-      "eval_steps_per_second": 12.263,
+      "epoch": 23.0,
+      "eval_accuracy": 0.5963589743589743,
+      "eval_loss": 1.8427150249481201,
+      "eval_runtime": 4.8431,
+      "eval_samples_per_second": 103.239,
+      "eval_steps_per_second": 13.008,
       "step": 4312
     },
     {
-      "epoch": 22.997333333333334,
-      "eval_exact_match": 13.6,
-      "eval_f1": 21.84761904761906,
+      "epoch": 23.0,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.13841036414565827,
       "step": 4312
     },
     {
-      "epoch": 23.466666666666665,
-      "grad_norm": 0.4574826955795288,
+      "epoch": 23.47,
       "learning_rate": 0.0003,
-      "loss": 0.4753,
+      "loss": 0.4186,
       "step": 4400
     },
     {
       "epoch": 24.0,
-      "grad_norm": 0.5842512249946594,
       "learning_rate": 0.0003,
-      "loss": 0.492,
+      "loss": 0.4365,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.5552307692307692,
-      "eval_loss": 2.3685286045074463,
-      "eval_runtime": 5.2131,
-      "eval_samples_per_second": 95.913,
-      "eval_steps_per_second": 12.085,
+      "eval_accuracy": 0.5993333333333334,
+      "eval_loss": 1.8092677593231201,
+      "eval_runtime": 4.206,
+      "eval_samples_per_second": 118.879,
+      "eval_steps_per_second": 14.979,
       "step": 4500
     },
     {
       "epoch": 24.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.71984126984128,
+      "eval_exact_match": 1.2,
+      "eval_f1": 4.495396065922379,
       "step": 4500
     },
     {
-      "epoch": 24.533333333333335,
-      "grad_norm": 0.41557422280311584,
+      "epoch": 24.53,
       "learning_rate": 0.0003,
-      "loss": 0.4728,
+      "loss": 0.4153,
       "step": 4600
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_accuracy": 0.5543076923076923,
-      "eval_loss": 2.37522554397583,
-      "eval_runtime": 5.8823,
-      "eval_samples_per_second": 85.001,
-      "eval_steps_per_second": 10.71,
+      "epoch": 25.0,
+      "eval_accuracy": 0.6022051282051282,
+      "eval_loss": 1.841846227645874,
+      "eval_runtime": 4.618,
+      "eval_samples_per_second": 108.271,
+      "eval_steps_per_second": 13.642,
       "step": 4687
     },
     {
-      "epoch": 24.997333333333334,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.936666666666675,
+      "epoch": 25.0,
+      "eval_exact_match": 1.6,
+      "eval_f1": 2.857142857142857,
       "step": 4687
     },
     {
-      "epoch": 25.066666666666666,
-      "grad_norm": 0.46733421087265015,
+      "epoch": 25.07,
       "learning_rate": 0.0003,
-      "loss": 0.4864,
+      "loss": 0.4275,
       "step": 4700
     },
     {
       "epoch": 25.6,
-      "grad_norm": 0.3511659502983093,
       "learning_rate": 0.0003,
-      "loss": 0.4753,
+      "loss": 0.4099,
       "step": 4800
     },
     {
       "epoch": 26.0,
-      "eval_accuracy": 0.5556923076923077,
-      "eval_loss": 2.3275949954986572,
-      "eval_runtime": 5.2145,
-      "eval_samples_per_second": 95.887,
-      "eval_steps_per_second": 12.082,
+      "eval_accuracy": 0.6001538461538461,
+      "eval_loss": 1.974199652671814,
+      "eval_runtime": 4.4307,
+      "eval_samples_per_second": 112.848,
+      "eval_steps_per_second": 14.219,
       "step": 4875
     },
     {
       "epoch": 26.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 22.252857142857145,
+      "eval_exact_match": 1.8,
+      "eval_f1": 2.9238095238095236,
       "step": 4875
     },
     {
-      "epoch": 26.133333333333333,
-      "grad_norm": 0.7332673072814941,
+      "epoch": 26.13,
       "learning_rate": 0.0003,
-      "loss": 0.486,
+      "loss": 0.4165,
       "step": 4900
     },
     {
-      "epoch": 26.666666666666668,
-      "grad_norm": 0.49349483847618103,
+      "epoch": 26.67,
       "learning_rate": 0.0003,
-      "loss": 0.4788,
+      "loss": 0.4101,
       "step": 5000
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_accuracy": 0.5541538461538461,
-      "eval_loss": 2.4251797199249268,
-      "eval_runtime": 5.8394,
-      "eval_samples_per_second": 85.625,
-      "eval_steps_per_second": 10.789,
+      "epoch": 27.0,
+      "eval_accuracy": 0.5987179487179487,
+      "eval_loss": 1.968170166015625,
+      "eval_runtime": 4.4099,
+      "eval_samples_per_second": 113.38,
+      "eval_steps_per_second": 14.286,
       "step": 5062
     },
     {
-      "epoch": 26.997333333333334,
-      "eval_exact_match": 15.0,
-      "eval_f1": 21.930952380952384,
+      "epoch": 27.0,
+      "eval_exact_match": 1.4,
+      "eval_f1": 2.3733333333333335,
       "step": 5062
     },
     {
       "epoch": 27.2,
-      "grad_norm": 0.5462525486946106,
       "learning_rate": 0.0003,
-      "loss": 0.482,
+      "loss": 0.4135,
       "step": 5100
     },
     {
-      "epoch": 27.733333333333334,
-      "grad_norm": 0.3605049252510071,
+      "epoch": 27.73,
       "learning_rate": 0.0003,
-      "loss": 0.4812,
+      "loss": 0.4122,
       "step": 5200
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.555076923076923,
-      "eval_loss": 2.381244659423828,
-      "eval_runtime": 5.142,
-      "eval_samples_per_second": 97.238,
-      "eval_steps_per_second": 12.252,
+      "eval_accuracy": 0.5991282051282051,
+      "eval_loss": 1.9906134605407715,
+      "eval_runtime": 4.1058,
+      "eval_samples_per_second": 121.778,
+      "eval_steps_per_second": 15.344,
       "step": 5250
     },
     {
       "epoch": 28.0,
-      "eval_exact_match": 14.4,
-      "eval_f1": 21.645238095238103,
+      "eval_exact_match": 0.0,
+      "eval_f1": 0.22539784029553853,
       "step": 5250
     }
   ],
   "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": 9.143294121685811e+16,
-  "train_batch_size": 1,
+  "total_flos": 1.9587996452284006e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-5250/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a48aaa718d03f9c3a1cf4e80dbff36330a05dd346d42a3b722f1ae2626aabe3c
-size 5304
+oid sha256:79a83a5d44cad230e338c4f4f6393fbd052223dad36807daa3baa278ceb8a412
+size 4728