Training in progress, epoch 16, checkpoint

checkpoint-3187/README.md

@@ -0,0 +1,9 @@
+---
+library_name: peft
+---
+## Training procedure
+
+### Framework versions
+
+
+- PEFT 0.5.0

checkpoint-3187/adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "Qwen/Qwen1.5-4B",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 16,
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "k_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "up_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-3187/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:249ab3cbcf070bfcfdf4485ec80baeeac932f1fa75a6908bb3b6adaf9e20b56f
+size 112130416

checkpoint-3187/added_tokens.json

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

checkpoint-3187/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f07a3b18c0ba1e684b2a5a203501f7c51ef0fc8187f985d06dabdb5f2d2e2e44
+size 224537202

checkpoint-3187/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b564a20b762c98ace2f43c2adbdcbed237c1548a8746d8cedff91c972a52659
+size 15024

checkpoint-3187/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:774a8a320fd9c2e71fdd237ac5879a747917e3ce2f14c71912cf47759ec65f73
+size 15024

checkpoint-3187/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd1be5a0f70af14a6b45747deb305c9e7c352db5e1bdb7998d8af0239530e8ef
+size 15024

checkpoint-3187/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4156244c69a0cf966ac120ec7eb209a18c2aa580593988ea3413287b376bbee8
+size 15024

checkpoint-3187/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31117d5ebc069331276648175d872c042616f6270576c899d59ce3fcd279dec8
+size 1064

checkpoint-3187/special_tokens_map.json

@@ -0,0 +1,14 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<|endoftext|>"
+}

checkpoint-3187/tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "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 %}",
+  "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
+}

checkpoint-3187/trainer_state.json

@@ -0,0 +1,573 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 16.997333333333334,
+  "eval_steps": 500,
+  "global_step": 3187,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.271308958530426,
+      "learning_rate": 5e-05,
+      "loss": 1.8262,
+      "step": 100
+    },
+    {
+      "epoch": 0.9973333333333333,
+      "eval_accuracy": 0.6047174887892377,
+      "eval_loss": 1.7027028799057007,
+      "eval_runtime": 6.6338,
+      "eval_samples_per_second": 75.372,
+      "eval_steps_per_second": 9.497,
+      "step": 187
+    },
+    {
+      "epoch": 0.9973333333333333,
+      "eval_exact_match": 14.6,
+      "eval_f1": 23.665714285714298,
+      "eval_qa_bleu": 10.24028390049623,
+      "eval_qa_exact_match": 0.114,
+      "eval_recite_bleu": 12.525534079250063,
+      "eval_recite_exact_match": 0.0,
+      "step": 187
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.301651269197464,
+      "learning_rate": 5e-05,
+      "loss": 1.7424,
+      "step": 200
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.35949045419692993,
+      "learning_rate": 5e-05,
+      "loss": 1.6978,
+      "step": 300
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6068699551569506,
+      "eval_loss": 1.6845285892486572,
+      "eval_runtime": 5.7524,
+      "eval_samples_per_second": 86.92,
+      "eval_steps_per_second": 10.952,
+      "step": 375
+    },
+    {
+      "epoch": 2.0,
+      "eval_exact_match": 14.6,
+      "eval_f1": 23.74190476190477,
+      "eval_qa_bleu": 9.70432698897927,
+      "eval_qa_exact_match": 0.114,
+      "eval_recite_bleu": 13.014822566317982,
+      "eval_recite_exact_match": 0.006,
+      "step": 375
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.41301965713500977,
+      "learning_rate": 5e-05,
+      "loss": 1.694,
+      "step": 400
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5057485699653625,
+      "learning_rate": 5e-05,
+      "loss": 1.6324,
+      "step": 500
+    },
+    {
+      "epoch": 2.997333333333333,
+      "eval_accuracy": 0.6079551569506726,
+      "eval_loss": 1.6851012706756592,
+      "eval_runtime": 5.5314,
+      "eval_samples_per_second": 90.394,
+      "eval_steps_per_second": 11.39,
+      "step": 562
+    },
+    {
+      "epoch": 2.997333333333333,
+      "eval_exact_match": 14.2,
+      "eval_f1": 23.29190476190477,
+      "eval_qa_bleu": 10.724307586340888,
+      "eval_qa_exact_match": 0.104,
+      "eval_recite_bleu": 12.83531066630853,
+      "eval_recite_exact_match": 0.004,
+      "step": 562
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.6363334655761719,
+      "learning_rate": 5e-05,
+      "loss": 1.6063,
+      "step": 600
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 0.6466453671455383,
+      "learning_rate": 5e-05,
+      "loss": 1.5562,
+      "step": 700
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.6071121076233184,
+      "eval_loss": 1.6969172954559326,
+      "eval_runtime": 6.2612,
+      "eval_samples_per_second": 79.857,
+      "eval_steps_per_second": 10.062,
+      "step": 750
+    },
+    {
+      "epoch": 4.0,
+      "eval_exact_match": 14.0,
+      "eval_f1": 21.34333333333334,
+      "eval_qa_bleu": 10.250340106635626,
+      "eval_qa_exact_match": 0.084,
+      "eval_recite_bleu": 12.741830585824024,
+      "eval_recite_exact_match": 0.004,
+      "step": 750
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 0.8162885904312134,
+      "learning_rate": 5e-05,
+      "loss": 1.5052,
+      "step": 800
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.8799439072608948,
+      "learning_rate": 5e-05,
+      "loss": 1.469,
+      "step": 900
+    },
+    {
+      "epoch": 4.997333333333334,
+      "eval_accuracy": 0.6059282511210762,
+      "eval_loss": 1.7216957807540894,
+      "eval_runtime": 6.3012,
+      "eval_samples_per_second": 79.35,
+      "eval_steps_per_second": 9.998,
+      "step": 937
+    },
+    {
+      "epoch": 4.997333333333334,
+      "eval_exact_match": 12.6,
+      "eval_f1": 21.041428571428575,
+      "eval_qa_bleu": 8.239984727774104,
+      "eval_qa_exact_match": 0.094,
+      "eval_recite_bleu": 12.327373697720192,
+      "eval_recite_exact_match": 0.002,
+      "step": 937
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 0.9834449291229248,
+      "learning_rate": 5e-05,
+      "loss": 1.3845,
+      "step": 1000
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 0.9377098679542542,
+      "learning_rate": 5e-05,
+      "loss": 1.3927,
+      "step": 1100
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.6047354260089686,
+      "eval_loss": 1.7564114332199097,
+      "eval_runtime": 6.3447,
+      "eval_samples_per_second": 78.806,
+      "eval_steps_per_second": 9.93,
+      "step": 1125
+    },
+    {
+      "epoch": 6.0,
+      "eval_exact_match": 13.4,
+      "eval_f1": 21.70317460317461,
+      "eval_qa_bleu": 10.522554916579404,
+      "eval_qa_exact_match": 0.104,
+      "eval_recite_bleu": 12.383604385812577,
+      "eval_recite_exact_match": 0.006,
+      "step": 1125
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 1.1172072887420654,
+      "learning_rate": 5e-05,
+      "loss": 1.2924,
+      "step": 1200
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 1.0980473756790161,
+      "learning_rate": 5e-05,
+      "loss": 1.286,
+      "step": 1300
+    },
+    {
+      "epoch": 6.997333333333334,
+      "eval_accuracy": 0.6026995515695067,
+      "eval_loss": 1.7977122068405151,
+      "eval_runtime": 5.9021,
+      "eval_samples_per_second": 84.716,
+      "eval_steps_per_second": 10.674,
+      "step": 1312
+    },
+    {
+      "epoch": 6.997333333333334,
+      "eval_exact_match": 12.0,
+      "eval_f1": 20.07142857142858,
+      "eval_qa_bleu": 9.89685444999411,
+      "eval_qa_exact_match": 0.102,
+      "eval_recite_bleu": 12.555118604776576,
+      "eval_recite_exact_match": 0.008,
+      "step": 1312
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 1.1534159183502197,
+      "learning_rate": 5e-05,
+      "loss": 1.1956,
+      "step": 1400
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 1.1209707260131836,
+      "learning_rate": 5e-05,
+      "loss": 1.1891,
+      "step": 1500
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.6010403587443947,
+      "eval_loss": 1.840914249420166,
+      "eval_runtime": 6.3016,
+      "eval_samples_per_second": 79.345,
+      "eval_steps_per_second": 9.997,
+      "step": 1500
+    },
+    {
+      "epoch": 8.0,
+      "eval_exact_match": 12.8,
+      "eval_f1": 21.598730158730167,
+      "eval_qa_bleu": 9.863863368837224,
+      "eval_qa_exact_match": 0.104,
+      "eval_recite_bleu": 12.09730367063489,
+      "eval_recite_exact_match": 0.01,
+      "step": 1500
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 1.3646217584609985,
+      "learning_rate": 5e-05,
+      "loss": 1.0861,
+      "step": 1600
+    },
+    {
+      "epoch": 8.997333333333334,
+      "eval_accuracy": 0.5981255605381166,
+      "eval_loss": 1.9223005771636963,
+      "eval_runtime": 5.5092,
+      "eval_samples_per_second": 90.757,
+      "eval_steps_per_second": 11.435,
+      "step": 1687
+    },
+    {
+      "epoch": 8.997333333333334,
+      "eval_exact_match": 14.8,
+      "eval_f1": 22.160000000000004,
+      "eval_qa_bleu": 8.357776258354589,
+      "eval_qa_exact_match": 0.116,
+      "eval_recite_bleu": 12.76267532884902,
+      "eval_recite_exact_match": 0.01,
+      "step": 1687
+    },
+    {
+      "epoch": 9.066666666666666,
+      "grad_norm": 1.6948473453521729,
+      "learning_rate": 5e-05,
+      "loss": 1.0935,
+      "step": 1700
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 1.6024364233016968,
+      "learning_rate": 5e-05,
+      "loss": 1.0127,
+      "step": 1800
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.5973632286995516,
+      "eval_loss": 1.9414385557174683,
+      "eval_runtime": 5.5824,
+      "eval_samples_per_second": 89.568,
+      "eval_steps_per_second": 11.286,
+      "step": 1875
+    },
+    {
+      "epoch": 10.0,
+      "eval_exact_match": 12.4,
+      "eval_f1": 20.75746031746032,
+      "eval_qa_bleu": 8.760424886550503,
+      "eval_qa_exact_match": 0.088,
+      "eval_recite_bleu": 12.754935192029142,
+      "eval_recite_exact_match": 0.016,
+      "step": 1875
+    },
+    {
+      "epoch": 10.133333333333333,
+      "grad_norm": 1.6970032453536987,
+      "learning_rate": 5e-05,
+      "loss": 0.9788,
+      "step": 1900
+    },
+    {
+      "epoch": 10.666666666666666,
+      "grad_norm": 1.559332251548767,
+      "learning_rate": 5e-05,
+      "loss": 0.926,
+      "step": 2000
+    },
+    {
+      "epoch": 10.997333333333334,
+      "eval_accuracy": 0.5945201793721973,
+      "eval_loss": 2.0455663204193115,
+      "eval_runtime": 5.5259,
+      "eval_samples_per_second": 90.484,
+      "eval_steps_per_second": 11.401,
+      "step": 2062
+    },
+    {
+      "epoch": 10.997333333333334,
+      "eval_exact_match": 13.0,
+      "eval_f1": 20.56190476190476,
+      "eval_qa_bleu": 11.437402190353293,
+      "eval_qa_exact_match": 0.106,
+      "eval_recite_bleu": 13.600786696850818,
+      "eval_recite_exact_match": 0.022,
+      "step": 2062
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 1.667855143547058,
+      "learning_rate": 5e-05,
+      "loss": 0.8872,
+      "step": 2100
+    },
+    {
+      "epoch": 11.733333333333333,
+      "grad_norm": 1.6040037870407104,
+      "learning_rate": 5e-05,
+      "loss": 0.8347,
+      "step": 2200
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.5923946188340807,
+      "eval_loss": 2.116677761077881,
+      "eval_runtime": 6.4038,
+      "eval_samples_per_second": 78.078,
+      "eval_steps_per_second": 9.838,
+      "step": 2250
+    },
+    {
+      "epoch": 12.0,
+      "eval_exact_match": 13.6,
+      "eval_f1": 20.470981240981256,
+      "eval_qa_bleu": 8.968260765274712,
+      "eval_qa_exact_match": 0.106,
+      "eval_recite_bleu": 12.793280320213494,
+      "eval_recite_exact_match": 0.014,
+      "step": 2250
+    },
+    {
+      "epoch": 12.266666666666667,
+      "grad_norm": 1.794092059135437,
+      "learning_rate": 5e-05,
+      "loss": 0.8032,
+      "step": 2300
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 1.8774526119232178,
+      "learning_rate": 5e-05,
+      "loss": 0.7736,
+      "step": 2400
+    },
+    {
+      "epoch": 12.997333333333334,
+      "eval_accuracy": 0.5902690582959641,
+      "eval_loss": 2.179492950439453,
+      "eval_runtime": 5.9945,
+      "eval_samples_per_second": 83.409,
+      "eval_steps_per_second": 10.51,
+      "step": 2437
+    },
+    {
+      "epoch": 12.997333333333334,
+      "eval_exact_match": 12.8,
+      "eval_f1": 20.780476190476197,
+      "eval_qa_bleu": 11.176823944321667,
+      "eval_qa_exact_match": 0.094,
+      "eval_recite_bleu": 13.292972390103905,
+      "eval_recite_exact_match": 0.02,
+      "step": 2437
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 1.9102168083190918,
+      "learning_rate": 5e-05,
+      "loss": 0.7168,
+      "step": 2500
+    },
+    {
+      "epoch": 13.866666666666667,
+      "grad_norm": 1.9183365106582642,
+      "learning_rate": 5e-05,
+      "loss": 0.6903,
+      "step": 2600
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.5875874439461883,
+      "eval_loss": 2.274404287338257,
+      "eval_runtime": 6.2502,
+      "eval_samples_per_second": 79.998,
+      "eval_steps_per_second": 10.08,
+      "step": 2625
+    },
+    {
+      "epoch": 14.0,
+      "eval_exact_match": 11.2,
+      "eval_f1": 18.26571428571429,
+      "eval_qa_bleu": 10.347091943612007,
+      "eval_qa_exact_match": 0.082,
+      "eval_recite_bleu": 12.720454352082847,
+      "eval_recite_exact_match": 0.014,
+      "step": 2625
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 2.326387405395508,
+      "learning_rate": 5e-05,
+      "loss": 0.6312,
+      "step": 2700
+    },
+    {
+      "epoch": 14.933333333333334,
+      "grad_norm": 2.1555840969085693,
+      "learning_rate": 5e-05,
+      "loss": 0.6267,
+      "step": 2800
+    },
+    {
+      "epoch": 14.997333333333334,
+      "eval_accuracy": 0.5870582959641255,
+      "eval_loss": 2.338937997817993,
+      "eval_runtime": 6.643,
+      "eval_samples_per_second": 75.267,
+      "eval_steps_per_second": 9.484,
+      "step": 2812
+    },
+    {
+      "epoch": 14.997333333333334,
+      "eval_exact_match": 10.6,
+      "eval_f1": 18.311471861471865,
+      "eval_qa_bleu": 9.56012066081651,
+      "eval_qa_exact_match": 0.086,
+      "eval_recite_bleu": 12.790292247765784,
+      "eval_recite_exact_match": 0.022,
+      "step": 2812
+    },
+    {
+      "epoch": 15.466666666666667,
+      "grad_norm": 1.8364830017089844,
+      "learning_rate": 5e-05,
+      "loss": 0.5582,
+      "step": 2900
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 1.9223560094833374,
+      "learning_rate": 5e-05,
+      "loss": 0.5673,
+      "step": 3000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.585067264573991,
+      "eval_loss": 2.4404428005218506,
+      "eval_runtime": 6.2254,
+      "eval_samples_per_second": 80.316,
+      "eval_steps_per_second": 10.12,
+      "step": 3000
+    },
+    {
+      "epoch": 16.0,
+      "eval_exact_match": 11.4,
+      "eval_f1": 18.553809523809527,
+      "eval_qa_bleu": 7.054735448183399,
+      "eval_qa_exact_match": 0.088,
+      "eval_recite_bleu": 12.41180854100743,
+      "eval_recite_exact_match": 0.018,
+      "step": 3000
+    },
+    {
+      "epoch": 16.533333333333335,
+      "grad_norm": 2.403259515762329,
+      "learning_rate": 5e-05,
+      "loss": 0.4886,
+      "step": 3100
+    },
+    {
+      "epoch": 16.997333333333334,
+      "eval_accuracy": 0.5844035874439462,
+      "eval_loss": 2.5225181579589844,
+      "eval_runtime": 5.9008,
+      "eval_samples_per_second": 84.734,
+      "eval_steps_per_second": 10.677,
+      "step": 3187
+    },
+    {
+      "epoch": 16.997333333333334,
+      "eval_exact_match": 10.6,
+      "eval_f1": 17.309567099567108,
+      "eval_qa_bleu": 7.6942420610670945,
+      "eval_qa_exact_match": 0.086,
+      "eval_recite_bleu": 11.974979252669918,
+      "eval_recite_exact_match": 0.016,
+      "step": 3187
+    }
+  ],
+  "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": 3.2107127369210266e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-3187/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01f9f0dc44c57fb44afb03320f4d02bf48066009e3bbcd96d8e4377b0171c5a5
+size 5304