Training in progress, epoch 9, checkpoint

checkpoint-2250/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-2250/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16ca3a6b261612f6622141a8025204c72e5c06c12160eaacab5f76cfa5a4e448
+size 143269386

checkpoint-2250/added_tokens.json

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

checkpoint-2250/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d71b0bf5f5a2058160d9c2ffa6d4508293fc979ae0fe95510550a502eedd28c9
-size 224537202
+oid sha256:41e1b27220deeb0f5a9e5be1d57a8f295ad3625bd735243d73bd7c47e6fc056a
+size 286585234

checkpoint-2250/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f69845742753bfd16862945ac4841bfe130548cfa92444266553293989378c61
+oid sha256:95f6dece1eb98c35f7fca80586bdc6ca173f070005be7a51d9f837711026687c
 size 15024

checkpoint-2250/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:61904efaf3b694e82ea7ee6221b63cd5cbc8ed7686c469ae2f3e2c6e77bd1569
+oid sha256:e60231ded3a0a54952d9e5d5e5a1c82c00f4bec47b424aadbfecbd4119479829
 size 15024

checkpoint-2250/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:14b2c09077123fe5c74b8de3fc87eb6a4939c7dca228929d16e03134275c9f55
+oid sha256:21e3a02087037c360302a3b51d3768d20b360fcc96513f2882e266c116f093b1
 size 15024

checkpoint-2250/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0cdd67374c140d2e465164e9e0e9a491e3547b1fd2a3e02ad0292b2d3e4b06c6
+oid sha256:9c5a0d44c1a02ba671ea4193ed0aee228ae08982de0a6724058919f783883731
 size 15024

checkpoint-2250/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-2250/tokenizer.model

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

checkpoint-2250/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-2250/trainer_state.json

@@ -10,313 +10,277 @@
   "log_history": [
     {
       "epoch": 0.4,
-      "grad_norm": 0.6251392960548401,
       "learning_rate": 5e-05,
-      "loss": 2.427,
+      "loss": 2.1914,
       "step": 100
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.6590011715888977,
       "learning_rate": 5e-05,
-      "loss": 2.2748,
+      "loss": 1.798,
       "step": 200
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5142539682539683,
-      "eval_loss": 2.336690902709961,
-      "eval_runtime": 5.4733,
-      "eval_samples_per_second": 91.353,
-      "eval_steps_per_second": 11.51,
+      "eval_accuracy": 0.6067341772151899,
+      "eval_loss": 1.8213441371917725,
+      "eval_runtime": 4.8116,
+      "eval_samples_per_second": 103.916,
+      "eval_steps_per_second": 13.093,
       "step": 250
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 16.8,
-      "eval_f1": 24.850246583481884,
+      "eval_exact_match": 25.4,
+      "eval_f1": 35.50992875545509,
       "step": 250
     },
     {
       "epoch": 1.2,
-      "grad_norm": 0.8194239139556885,
       "learning_rate": 5e-05,
-      "loss": 2.2322,
+      "loss": 1.7534,
       "step": 300
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.9867294430732727,
       "learning_rate": 5e-05,
-      "loss": 2.1904,
+      "loss": 1.7163,
       "step": 400
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.9660400152206421,
       "learning_rate": 5e-05,
-      "loss": 2.1657,
+      "loss": 1.7,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5168253968253969,
-      "eval_loss": 2.323688507080078,
-      "eval_runtime": 5.1683,
-      "eval_samples_per_second": 96.744,
-      "eval_steps_per_second": 12.19,
+      "eval_accuracy": 0.6077215189873417,
+      "eval_loss": 1.8046445846557617,
+      "eval_runtime": 5.1228,
+      "eval_samples_per_second": 97.603,
+      "eval_steps_per_second": 12.298,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 24.620036075036083,
+      "eval_exact_match": 25.8,
+      "eval_f1": 35.76454372424963,
       "step": 500
     },
     {
       "epoch": 2.4,
-      "grad_norm": 1.281128168106079,
       "learning_rate": 5e-05,
-      "loss": 2.0251,
+      "loss": 1.5854,
       "step": 600
     },
     {
       "epoch": 2.8,
-      "grad_norm": 1.3365116119384766,
       "learning_rate": 5e-05,
-      "loss": 2.0158,
+      "loss": 1.5869,
       "step": 700
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.5156507936507937,
-      "eval_loss": 2.3560824394226074,
-      "eval_runtime": 5.4675,
-      "eval_samples_per_second": 91.449,
-      "eval_steps_per_second": 11.523,
+      "eval_accuracy": 0.6071392405063291,
+      "eval_loss": 1.829284906387329,
+      "eval_runtime": 4.5692,
+      "eval_samples_per_second": 109.429,
+      "eval_steps_per_second": 13.788,
       "step": 750
     },
     {
       "epoch": 3.0,
-      "eval_exact_match": 19.0,
-      "eval_f1": 27.242790688843332,
+      "eval_exact_match": 24.4,
+      "eval_f1": 34.46831890331891,
       "step": 750
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.7840828895568848,
       "learning_rate": 5e-05,
-      "loss": 1.9394,
+      "loss": 1.5039,
       "step": 800
     },
     {
       "epoch": 3.6,
-      "grad_norm": 1.8030190467834473,
       "learning_rate": 5e-05,
-      "loss": 1.8502,
+      "loss": 1.4165,
       "step": 900
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.9682222604751587,
       "learning_rate": 5e-05,
-      "loss": 1.8682,
+      "loss": 1.4349,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5131746031746032,
-      "eval_loss": 2.4071946144104004,
-      "eval_runtime": 5.2825,
-      "eval_samples_per_second": 94.653,
-      "eval_steps_per_second": 11.926,
+      "eval_accuracy": 0.6042784810126582,
+      "eval_loss": 1.89736008644104,
+      "eval_runtime": 4.5698,
+      "eval_samples_per_second": 109.415,
+      "eval_steps_per_second": 13.786,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 17.6,
-      "eval_f1": 26.010988455988468,
+      "eval_exact_match": 22.8,
+      "eval_f1": 32.991320346320364,
       "step": 1000
     },
     {
       "epoch": 4.4,
-      "grad_norm": 2.138505220413208,
       "learning_rate": 5e-05,
-      "loss": 1.6795,
+      "loss": 1.2624,
       "step": 1100
     },
     {
       "epoch": 4.8,
-      "grad_norm": 2.3212926387786865,
       "learning_rate": 5e-05,
-      "loss": 1.7036,
+      "loss": 1.3111,
       "step": 1200
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.5099365079365079,
-      "eval_loss": 2.494770050048828,
-      "eval_runtime": 5.1799,
-      "eval_samples_per_second": 96.528,
-      "eval_steps_per_second": 12.162,
+      "eval_accuracy": 0.601493670886076,
+      "eval_loss": 1.9769378900527954,
+      "eval_runtime": 4.7168,
+      "eval_samples_per_second": 106.005,
+      "eval_steps_per_second": 13.357,
       "step": 1250
     },
     {
       "epoch": 5.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 25.84752525252526,
+      "eval_exact_match": 22.6,
+      "eval_f1": 32.46891774891776,
       "step": 1250
     },
     {
       "epoch": 5.2,
-      "grad_norm": 2.4878485202789307,
       "learning_rate": 5e-05,
-      "loss": 1.6305,
+      "loss": 1.226,
       "step": 1300
     },
     {
       "epoch": 5.6,
-      "grad_norm": 2.599329710006714,
       "learning_rate": 5e-05,
-      "loss": 1.5477,
+      "loss": 1.1608,
       "step": 1400
     },
     {
       "epoch": 6.0,
-      "grad_norm": 2.5542759895324707,
       "learning_rate": 5e-05,
-      "loss": 1.5708,
+      "loss": 1.197,
       "step": 1500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5065079365079365,
-      "eval_loss": 2.5797038078308105,
-      "eval_runtime": 5.6827,
-      "eval_samples_per_second": 87.986,
-      "eval_steps_per_second": 11.086,
+      "eval_accuracy": 0.599240506329114,
+      "eval_loss": 2.0634872913360596,
+      "eval_runtime": 4.4047,
+      "eval_samples_per_second": 113.515,
+      "eval_steps_per_second": 14.303,
       "step": 1500
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 17.0,
-      "eval_f1": 25.310858585858586,
+      "eval_exact_match": 21.4,
+      "eval_f1": 29.995800865800874,
       "step": 1500
     },
     {
       "epoch": 6.4,
-      "grad_norm": 3.1968846321105957,
       "learning_rate": 5e-05,
-      "loss": 1.3896,
+      "loss": 1.0336,
       "step": 1600
     },
     {
       "epoch": 6.8,
-      "grad_norm": 3.8637375831604004,
       "learning_rate": 5e-05,
-      "loss": 1.421,
+      "loss": 1.0729,
       "step": 1700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.5038412698412699,
-      "eval_loss": 2.6977317333221436,
-      "eval_runtime": 5.3312,
-      "eval_samples_per_second": 93.787,
-      "eval_steps_per_second": 11.817,
+      "eval_accuracy": 0.5975443037974684,
+      "eval_loss": 2.1523237228393555,
+      "eval_runtime": 4.4025,
+      "eval_samples_per_second": 113.571,
+      "eval_steps_per_second": 14.31,
       "step": 1750
     },
     {
       "epoch": 7.0,
-      "eval_exact_match": 16.0,
-      "eval_f1": 23.99441558441559,
+      "eval_exact_match": 20.8,
+      "eval_f1": 29.054994058229354,
       "step": 1750
     },
     {
       "epoch": 7.2,
-      "grad_norm": 3.468459367752075,
       "learning_rate": 5e-05,
-      "loss": 1.3366,
+      "loss": 1.0005,
       "step": 1800
     },
     {
       "epoch": 7.6,
-      "grad_norm": 3.5727667808532715,
       "learning_rate": 5e-05,
-      "loss": 1.2585,
+      "loss": 0.9437,
       "step": 1900
     },
     {
       "epoch": 8.0,
-      "grad_norm": 3.508258819580078,
       "learning_rate": 5e-05,
-      "loss": 1.296,
+      "loss": 0.9833,
       "step": 2000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.5013333333333333,
-      "eval_loss": 2.786928653717041,
-      "eval_runtime": 5.8398,
-      "eval_samples_per_second": 85.62,
-      "eval_steps_per_second": 10.788,
+      "eval_accuracy": 0.5947341772151898,
+      "eval_loss": 2.2639756202697754,
+      "eval_runtime": 4.7031,
+      "eval_samples_per_second": 106.312,
+      "eval_steps_per_second": 13.395,
       "step": 2000
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 22.542871572871572,
+      "eval_exact_match": 20.8,
+      "eval_f1": 28.71228732052262,
       "step": 2000
     },
     {
       "epoch": 8.4,
-      "grad_norm": 4.042112350463867,
       "learning_rate": 5e-05,
-      "loss": 1.1084,
+      "loss": 0.823,
       "step": 2100
     },
     {
       "epoch": 8.8,
-      "grad_norm": 3.7548155784606934,
       "learning_rate": 5e-05,
-      "loss": 1.1457,
+      "loss": 0.8672,
       "step": 2200
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.49692063492063493,
-      "eval_loss": 2.9215197563171387,
-      "eval_runtime": 5.1412,
-      "eval_samples_per_second": 97.254,
-      "eval_steps_per_second": 12.254,
+      "eval_accuracy": 0.5924050632911393,
+      "eval_loss": 2.3642578125,
+      "eval_runtime": 4.4051,
+      "eval_samples_per_second": 113.505,
+      "eval_steps_per_second": 14.302,
       "step": 2250
     },
     {
       "epoch": 9.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 21.59087301587302,
+      "eval_exact_match": 22.4,
+      "eval_f1": 30.14020120686362,
       "step": 2250
     }
   ],
   "logging_steps": 100,
   "max_steps": 12500,
-  "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": 5.82092363619369e+16,
-  "train_batch_size": 1,
+  "total_flos": 1.4384866119594803e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-2250/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ede3565a5658e9b8743d6efdd527fd23709623a130469c81f1191d07c33fbcb1
-size 5304
+oid sha256:ddd08619aa8955df0f518864a252643438869f6968a9f390d3026fbeb48e7ddf
+size 4728