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:cdf6e8d1676264a9738645ae97bf4401f6a25b25de4d2901dd4f15f751465836
+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 sha256:3e929ac3a072a2ba1ab1e7041f94eed46cc7ed3ce23b065f7d5b8e5a57881751
-size 224537202
+oid sha256:6d345dc38058c8c44c9d9f5ee267c6bd9b6929cca072988fe7442f9857962182
+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": 0.48472875356674194,
       "learning_rate": 0.0003,
-      "loss": 2.3219,
+      "loss": 1.9296,
       "step": 100
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.4583953619003296,
       "learning_rate": 0.0003,
-      "loss": 2.2487,
+      "loss": 1.7554,
       "step": 200
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5168253968253969,
-      "eval_loss": 2.311781167984009,
-      "eval_runtime": 5.4425,
-      "eval_samples_per_second": 91.87,
-      "eval_steps_per_second": 11.576,
+      "eval_accuracy": 0.6093164556962025,
+      "eval_loss": 1.7939746379852295,
+      "eval_runtime": 5.0177,
+      "eval_samples_per_second": 99.648,
+      "eval_steps_per_second": 12.556,
       "step": 250
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 18.6,
-      "eval_f1": 26.999121042356347,
+      "eval_exact_match": 26.0,
+      "eval_f1": 36.71136363636365,
       "step": 250
     },
     {
       "epoch": 1.2,
-      "grad_norm": 0.6610094904899597,
       "learning_rate": 0.0003,
-      "loss": 2.083,
+      "loss": 1.6394,
       "step": 300
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.7361629605293274,
       "learning_rate": 0.0003,
-      "loss": 1.9274,
+      "loss": 1.5315,
       "step": 400
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.6793610453605652,
       "learning_rate": 0.0003,
-      "loss": 1.9113,
+      "loss": 1.5248,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5153650793650794,
-      "eval_loss": 2.3749232292175293,
-      "eval_runtime": 5.2167,
-      "eval_samples_per_second": 95.846,
-      "eval_steps_per_second": 12.077,
+      "eval_accuracy": 0.6085063291139241,
+      "eval_loss": 1.8273799419403076,
+      "eval_runtime": 4.4023,
+      "eval_samples_per_second": 113.576,
+      "eval_steps_per_second": 14.311,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 18.2,
-      "eval_f1": 26.209704184704197,
+      "eval_exact_match": 24.4,
+      "eval_f1": 33.57723554223556,
       "step": 500
     },
     {
       "epoch": 2.4,
-      "grad_norm": 0.8843489289283752,
       "learning_rate": 0.0003,
-      "loss": 1.4602,
+      "loss": 1.1692,
       "step": 600
     },
     {
       "epoch": 2.8,
-      "grad_norm": 0.8380613923072815,
       "learning_rate": 0.0003,
-      "loss": 1.4924,
+      "loss": 1.2054,
       "step": 700
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.5098730158730159,
-      "eval_loss": 2.5289740562438965,
-      "eval_runtime": 5.3631,
-      "eval_samples_per_second": 93.23,
-      "eval_steps_per_second": 11.747,
+      "eval_accuracy": 0.6027088607594937,
+      "eval_loss": 1.9717934131622314,
+      "eval_runtime": 5.1273,
+      "eval_samples_per_second": 97.518,
+      "eval_steps_per_second": 12.287,
       "step": 750
     },
     {
       "epoch": 3.0,
-      "eval_exact_match": 16.6,
-      "eval_f1": 24.526153291153296,
+      "eval_exact_match": 20.8,
+      "eval_f1": 30.584243697478996,
       "step": 750
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.068247675895691,
       "learning_rate": 0.0003,
-      "loss": 1.2869,
+      "loss": 1.0252,
       "step": 800
     },
     {
       "epoch": 3.6,
-      "grad_norm": 1.041741967201233,
       "learning_rate": 0.0003,
-      "loss": 1.0749,
+      "loss": 0.8608,
       "step": 900
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.267968773841858,
       "learning_rate": 0.0003,
-      "loss": 1.1157,
+      "loss": 0.8989,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5038412698412699,
-      "eval_loss": 2.794942855834961,
-      "eval_runtime": 5.0088,
-      "eval_samples_per_second": 99.825,
-      "eval_steps_per_second": 12.578,
+      "eval_accuracy": 0.5986835443037974,
+      "eval_loss": 2.151914596557617,
+      "eval_runtime": 5.3325,
+      "eval_samples_per_second": 93.765,
+      "eval_steps_per_second": 11.814,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 16.4,
-      "eval_f1": 23.978561993561993,
+      "eval_exact_match": 21.0,
+      "eval_f1": 30.363229548229558,
       "step": 1000
     },
     {
       "epoch": 4.4,
-      "grad_norm": 1.1971005201339722,
       "learning_rate": 0.0003,
-      "loss": 0.7367,
+      "loss": 0.5842,
       "step": 1100
     },
     {
       "epoch": 4.8,
-      "grad_norm": 1.2383127212524414,
       "learning_rate": 0.0003,
-      "loss": 0.7814,
+      "loss": 0.6306,
       "step": 1200
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.49853968253968256,
-      "eval_loss": 3.0137181282043457,
-      "eval_runtime": 5.1368,
-      "eval_samples_per_second": 97.337,
-      "eval_steps_per_second": 12.264,
+      "eval_accuracy": 0.5960506329113924,
+      "eval_loss": 2.329284191131592,
+      "eval_runtime": 4.5585,
+      "eval_samples_per_second": 109.686,
+      "eval_steps_per_second": 13.82,
       "step": 1250
     },
     {
       "epoch": 5.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 21.880294705294705,
+      "eval_exact_match": 20.4,
+      "eval_f1": 30.02542624042625,
       "step": 1250
     },
     {
       "epoch": 5.2,
-      "grad_norm": 0.9558005928993225,
       "learning_rate": 0.0003,
-      "loss": 0.6786,
+      "loss": 0.5368,
       "step": 1300
     },
     {
       "epoch": 5.6,
-      "grad_norm": 1.1341361999511719,
       "learning_rate": 0.0003,
-      "loss": 0.5741,
+      "loss": 0.4495,
       "step": 1400
     },
     {
       "epoch": 6.0,
-      "grad_norm": 1.1375033855438232,
       "learning_rate": 0.0003,
-      "loss": 0.605,
+      "loss": 0.4712,
       "step": 1500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.49657142857142855,
-      "eval_loss": 3.220507860183716,
-      "eval_runtime": 5.2708,
-      "eval_samples_per_second": 94.863,
-      "eval_steps_per_second": 11.953,
+      "eval_accuracy": 0.5935696202531645,
+      "eval_loss": 2.5598793029785156,
+      "eval_runtime": 4.4094,
+      "eval_samples_per_second": 113.395,
+      "eval_steps_per_second": 14.288,
       "step": 1500
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 15.4,
-      "eval_f1": 22.281807987334307,
+      "eval_exact_match": 20.6,
+      "eval_f1": 30.108149073149082,
       "step": 1500
     },
     {
       "epoch": 6.4,
-      "grad_norm": 1.118528962135315,
       "learning_rate": 0.0003,
-      "loss": 0.4684,
+      "loss": 0.3625,
       "step": 1600
     },
     {
       "epoch": 6.8,
-      "grad_norm": 0.9311291575431824,
       "learning_rate": 0.0003,
-      "loss": 0.4968,
+      "loss": 0.3797,
       "step": 1700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.4946984126984127,
-      "eval_loss": 3.3760199546813965,
-      "eval_runtime": 5.0734,
-      "eval_samples_per_second": 98.554,
-      "eval_steps_per_second": 12.418,
+      "eval_accuracy": 0.5935696202531645,
+      "eval_loss": 2.732860803604126,
+      "eval_runtime": 5.2701,
+      "eval_samples_per_second": 94.876,
+      "eval_steps_per_second": 11.954,
       "step": 1750
     },
     {
       "epoch": 7.0,
-      "eval_exact_match": 14.8,
-      "eval_f1": 21.79480811586075,
+      "eval_exact_match": 19.6,
+      "eval_f1": 29.002709512709526,
       "step": 1750
     },
     {
       "epoch": 7.2,
-      "grad_norm": 1.0299904346466064,
       "learning_rate": 0.0003,
-      "loss": 0.4631,
+      "loss": 0.3582,
       "step": 1800
     },
     {
       "epoch": 7.6,
-      "grad_norm": 0.7518107295036316,
       "learning_rate": 0.0003,
-      "loss": 0.4411,
+      "loss": 0.3399,
       "step": 1900
     },
     {
       "epoch": 8.0,
-      "grad_norm": 1.1140515804290771,
       "learning_rate": 0.0003,
-      "loss": 0.4597,
+      "loss": 0.3527,
       "step": 2000
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.49234920634920637,
-      "eval_loss": 3.4634218215942383,
-      "eval_runtime": 5.8652,
-      "eval_samples_per_second": 85.249,
-      "eval_steps_per_second": 10.741,
+      "eval_accuracy": 0.5912911392405064,
+      "eval_loss": 2.8185083866119385,
+      "eval_runtime": 4.4105,
+      "eval_samples_per_second": 113.366,
+      "eval_steps_per_second": 14.284,
       "step": 2000
     },
     {
       "epoch": 8.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 21.617115196631318,
+      "eval_exact_match": 20.8,
+      "eval_f1": 29.191525696525694,
       "step": 2000
     },
     {
       "epoch": 8.4,
-      "grad_norm": 1.2363754510879517,
       "learning_rate": 0.0003,
-      "loss": 0.4026,
+      "loss": 0.3202,
       "step": 2100
     },
     {
       "epoch": 8.8,
-      "grad_norm": 0.8174892067909241,
       "learning_rate": 0.0003,
-      "loss": 0.4242,
+      "loss": 0.3314,
       "step": 2200
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.4955873015873016,
-      "eval_loss": 3.4379212856292725,
-      "eval_runtime": 5.214,
-      "eval_samples_per_second": 95.896,
-      "eval_steps_per_second": 12.083,
+      "eval_accuracy": 0.592,
+      "eval_loss": 2.824962854385376,
+      "eval_runtime": 4.5596,
+      "eval_samples_per_second": 109.66,
+      "eval_steps_per_second": 13.817,
       "step": 2250
     },
     {
       "epoch": 9.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 22.36426184926185,
+      "eval_exact_match": 21.0,
+      "eval_f1": 29.434571539571547,
       "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:dae73bb50794744b259c1eaaeb67f6ae0b1d0f39b0a32b5f1162fc82e3a5d6a6
-size 5304
+oid sha256:eef995e0a4a8da1af8ae275504624b1fcb2d18639b8c53b553ebed1c9a9b7fd7
+size 4728