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 sha256:688dda4ae67145ed736f559828cf7b7541c8d36471481e462ce4962f6975b92a
+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 sha256:0e11547a79ef0d428e6a6fa9cc5c8f0589902ddd32f4b4758938403e915fe0dd
+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 sha256:838aefe87888f08aaf4f1b22ea7029ebc788537ff5f89bf7b629a97d2b9b8fe5
+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": 0.2081739604473114,
+      "learning_rate": 0.0005,
+      "loss": 1.7461,
+      "step": 100
+    },
+    {
+      "epoch": 0.9973333333333333,
+      "eval_accuracy": 0.6093004484304932,
+      "eval_loss": 1.6695144176483154,
+      "eval_runtime": 6.0389,
+      "eval_samples_per_second": 82.796,
+      "eval_steps_per_second": 10.432,
+      "step": 187
+    },
+    {
+      "epoch": 0.9973333333333333,
+      "eval_exact_match": 15.8,
+      "eval_f1": 24.64418803418805,
+      "eval_qa_bleu": 12.64258947770803,
+      "eval_qa_exact_match": 0.116,
+      "eval_recite_bleu": 12.676034356605994,
+      "eval_recite_exact_match": 0.002,
+      "step": 187
+    },
+    {
+      "epoch": 1.0666666666666667,
+      "grad_norm": 0.2944607138633728,
+      "learning_rate": 0.0005,
+      "loss": 1.6853,
+      "step": 200
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.40688613057136536,
+      "learning_rate": 0.0005,
+      "loss": 1.4286,
+      "step": 300
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.6086547085201793,
+      "eval_loss": 1.6994106769561768,
+      "eval_runtime": 6.3284,
+      "eval_samples_per_second": 79.009,
+      "eval_steps_per_second": 9.955,
+      "step": 375
+    },
+    {
+      "epoch": 2.0,
+      "eval_exact_match": 13.6,
+      "eval_f1": 22.022914862914874,
+      "eval_qa_bleu": 11.191170451488254,
+      "eval_qa_exact_match": 0.098,
+      "eval_recite_bleu": 13.062440471503999,
+      "eval_recite_exact_match": 0.008,
+      "step": 375
+    },
+    {
+      "epoch": 2.1333333333333333,
+      "grad_norm": 0.4661327600479126,
+      "learning_rate": 0.0005,
+      "loss": 1.3391,
+      "step": 400
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.5310370326042175,
+      "learning_rate": 0.0005,
+      "loss": 1.0309,
+      "step": 500
+    },
+    {
+      "epoch": 2.997333333333333,
+      "eval_accuracy": 0.6053452914798206,
+      "eval_loss": 1.812237024307251,
+      "eval_runtime": 5.6113,
+      "eval_samples_per_second": 89.107,
+      "eval_steps_per_second": 11.227,
+      "step": 562
+    },
+    {
+      "epoch": 2.997333333333333,
+      "eval_exact_match": 12.8,
+      "eval_f1": 21.852727272727286,
+      "eval_qa_bleu": 11.536967687478375,
+      "eval_qa_exact_match": 0.1,
+      "eval_recite_bleu": 13.977710455440388,
+      "eval_recite_exact_match": 0.008,
+      "step": 562
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.5991603136062622,
+      "learning_rate": 0.0005,
+      "loss": 0.9162,
+      "step": 600
+    },
+    {
+      "epoch": 3.7333333333333334,
+      "grad_norm": 0.5921025276184082,
+      "learning_rate": 0.0005,
+      "loss": 0.7019,
+      "step": 700
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.5989417040358744,
+      "eval_loss": 1.974912166595459,
+      "eval_runtime": 6.3058,
+      "eval_samples_per_second": 79.292,
+      "eval_steps_per_second": 9.991,
+      "step": 750
+    },
+    {
+      "epoch": 4.0,
+      "eval_exact_match": 10.8,
+      "eval_f1": 18.718571428571437,
+      "eval_qa_bleu": 8.547340405175287,
+      "eval_qa_exact_match": 0.088,
+      "eval_recite_bleu": 13.524416242106161,
+      "eval_recite_exact_match": 0.018,
+      "step": 750
+    },
+    {
+      "epoch": 4.266666666666667,
+      "grad_norm": 0.6397003531455994,
+      "learning_rate": 0.0005,
+      "loss": 0.5811,
+      "step": 800
+    },
+    {
+      "epoch": 4.8,
+      "grad_norm": 0.6721498966217041,
+      "learning_rate": 0.0005,
+      "loss": 0.4634,
+      "step": 900
+    },
+    {
+      "epoch": 4.997333333333334,
+      "eval_accuracy": 0.5947892376681614,
+      "eval_loss": 2.1952831745147705,
+      "eval_runtime": 5.789,
+      "eval_samples_per_second": 86.37,
+      "eval_steps_per_second": 10.883,
+      "step": 937
+    },
+    {
+      "epoch": 4.997333333333334,
+      "eval_exact_match": 12.6,
+      "eval_f1": 20.56238095238097,
+      "eval_qa_bleu": 8.878599537366625,
+      "eval_qa_exact_match": 0.106,
+      "eval_recite_bleu": 14.97129401028483,
+      "eval_recite_exact_match": 0.018,
+      "step": 937
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 0.5803630352020264,
+      "learning_rate": 0.0005,
+      "loss": 0.3603,
+      "step": 1000
+    },
+    {
+      "epoch": 5.866666666666667,
+      "grad_norm": 0.7646645903587341,
+      "learning_rate": 0.0005,
+      "loss": 0.3066,
+      "step": 1100
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.5917309417040358,
+      "eval_loss": 2.372556209564209,
+      "eval_runtime": 6.4215,
+      "eval_samples_per_second": 77.863,
+      "eval_steps_per_second": 9.811,
+      "step": 1125
+    },
+    {
+      "epoch": 6.0,
+      "eval_exact_match": 14.0,
+      "eval_f1": 21.69952380952382,
+      "eval_qa_bleu": 11.132618008914505,
+      "eval_qa_exact_match": 0.104,
+      "eval_recite_bleu": 15.834212599206387,
+      "eval_recite_exact_match": 0.022,
+      "step": 1125
+    },
+    {
+      "epoch": 6.4,
+      "grad_norm": 0.5899696946144104,
+      "learning_rate": 0.0005,
+      "loss": 0.2279,
+      "step": 1200
+    },
+    {
+      "epoch": 6.933333333333334,
+      "grad_norm": 0.6446067690849304,
+      "learning_rate": 0.0005,
+      "loss": 0.2171,
+      "step": 1300
+    },
+    {
+      "epoch": 6.997333333333334,
+      "eval_accuracy": 0.5900089686098655,
+      "eval_loss": 2.5298404693603516,
+      "eval_runtime": 5.7188,
+      "eval_samples_per_second": 87.43,
+      "eval_steps_per_second": 11.016,
+      "step": 1312
+    },
+    {
+      "epoch": 6.997333333333334,
+      "eval_exact_match": 13.0,
+      "eval_f1": 20.44476190476192,
+      "eval_qa_bleu": 9.72146731421567,
+      "eval_qa_exact_match": 0.102,
+      "eval_recite_bleu": 16.053346195966608,
+      "eval_recite_exact_match": 0.028,
+      "step": 1312
+    },
+    {
+      "epoch": 7.466666666666667,
+      "grad_norm": 0.5204954743385315,
+      "learning_rate": 0.0005,
+      "loss": 0.165,
+      "step": 1400
+    },
+    {
+      "epoch": 8.0,
+      "grad_norm": 0.5783184170722961,
+      "learning_rate": 0.0005,
+      "loss": 0.1742,
+      "step": 1500
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.5902690582959641,
+      "eval_loss": 2.5951457023620605,
+      "eval_runtime": 6.1142,
+      "eval_samples_per_second": 81.777,
+      "eval_steps_per_second": 10.304,
+      "step": 1500
+    },
+    {
+      "epoch": 8.0,
+      "eval_exact_match": 14.0,
+      "eval_f1": 20.39317460317461,
+      "eval_qa_bleu": 12.13686711407266,
+      "eval_qa_exact_match": 0.114,
+      "eval_recite_bleu": 17.051674394691624,
+      "eval_recite_exact_match": 0.038,
+      "step": 1500
+    },
+    {
+      "epoch": 8.533333333333333,
+      "grad_norm": 0.4550940692424774,
+      "learning_rate": 0.0005,
+      "loss": 0.1376,
+      "step": 1600
+    },
+    {
+      "epoch": 8.997333333333334,
+      "eval_accuracy": 0.5896143497757848,
+      "eval_loss": 2.698397159576416,
+      "eval_runtime": 6.2934,
+      "eval_samples_per_second": 79.448,
+      "eval_steps_per_second": 10.01,
+      "step": 1687
+    },
+    {
+      "epoch": 8.997333333333334,
+      "eval_exact_match": 12.4,
+      "eval_f1": 18.647619047619052,
+      "eval_qa_bleu": 12.590236325584641,
+      "eval_qa_exact_match": 0.098,
+      "eval_recite_bleu": 16.764890988222987,
+      "eval_recite_exact_match": 0.04,
+      "step": 1687
+    },
+    {
+      "epoch": 9.066666666666666,
+      "grad_norm": 0.40663576126098633,
+      "learning_rate": 0.0005,
+      "loss": 0.1507,
+      "step": 1700
+    },
+    {
+      "epoch": 9.6,
+      "grad_norm": 0.4749195873737335,
+      "learning_rate": 0.0005,
+      "loss": 0.1325,
+      "step": 1800
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.588627802690583,
+      "eval_loss": 2.717050790786743,
+      "eval_runtime": 5.5734,
+      "eval_samples_per_second": 89.712,
+      "eval_steps_per_second": 11.304,
+      "step": 1875
+    },
+    {
+      "epoch": 10.0,
+      "eval_exact_match": 12.2,
+      "eval_f1": 19.749870129870143,
+      "eval_qa_bleu": 11.806166770872942,
+      "eval_qa_exact_match": 0.1,
+      "eval_recite_bleu": 17.357706445936504,
+      "eval_recite_exact_match": 0.048,
+      "step": 1875
+    },
+    {
+      "epoch": 10.133333333333333,
+      "grad_norm": 0.427946001291275,
+      "learning_rate": 0.0005,
+      "loss": 0.1428,
+      "step": 1900
+    },
+    {
+      "epoch": 10.666666666666666,
+      "grad_norm": 0.5029444694519043,
+      "learning_rate": 0.0005,
+      "loss": 0.133,
+      "step": 2000
+    },
+    {
+      "epoch": 10.997333333333334,
+      "eval_accuracy": 0.5878744394618834,
+      "eval_loss": 2.743403911590576,
+      "eval_runtime": 5.6067,
+      "eval_samples_per_second": 89.179,
+      "eval_steps_per_second": 11.237,
+      "step": 2062
+    },
+    {
+      "epoch": 10.997333333333334,
+      "eval_exact_match": 13.0,
+      "eval_f1": 20.467619047619053,
+      "eval_qa_bleu": 10.540745614436421,
+      "eval_qa_exact_match": 0.106,
+      "eval_recite_bleu": 17.42269517845921,
+      "eval_recite_exact_match": 0.036,
+      "step": 2062
+    },
+    {
+      "epoch": 11.2,
+      "grad_norm": 0.355397492647171,
+      "learning_rate": 0.0005,
+      "loss": 0.1402,
+      "step": 2100
+    },
+    {
+      "epoch": 11.733333333333333,
+      "grad_norm": 0.49750107526779175,
+      "learning_rate": 0.0005,
+      "loss": 0.1327,
+      "step": 2200
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.587354260089686,
+      "eval_loss": 2.760918378829956,
+      "eval_runtime": 6.2134,
+      "eval_samples_per_second": 80.471,
+      "eval_steps_per_second": 10.139,
+      "step": 2250
+    },
+    {
+      "epoch": 12.0,
+      "eval_exact_match": 13.6,
+      "eval_f1": 21.793809523809543,
+      "eval_qa_bleu": 9.887576652654696,
+      "eval_qa_exact_match": 0.104,
+      "eval_recite_bleu": 16.11733068263277,
+      "eval_recite_exact_match": 0.038,
+      "step": 2250
+    },
+    {
+      "epoch": 12.266666666666667,
+      "grad_norm": 0.4602002501487732,
+      "learning_rate": 0.0005,
+      "loss": 0.1388,
+      "step": 2300
+    },
+    {
+      "epoch": 12.8,
+      "grad_norm": 0.5736598372459412,
+      "learning_rate": 0.0005,
+      "loss": 0.1387,
+      "step": 2400
+    },
+    {
+      "epoch": 12.997333333333334,
+      "eval_accuracy": 0.5862152466367713,
+      "eval_loss": 2.7902321815490723,
+      "eval_runtime": 5.9757,
+      "eval_samples_per_second": 83.673,
+      "eval_steps_per_second": 10.543,
+      "step": 2437
+    },
+    {
+      "epoch": 12.997333333333334,
+      "eval_exact_match": 12.4,
+      "eval_f1": 19.832698412698424,
+      "eval_qa_bleu": 11.36986519306105,
+      "eval_qa_exact_match": 0.106,
+      "eval_recite_bleu": 16.527532176178507,
+      "eval_recite_exact_match": 0.032,
+      "step": 2437
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 0.40387147665023804,
+      "learning_rate": 0.0005,
+      "loss": 0.1349,
+      "step": 2500
+    },
+    {
+      "epoch": 13.866666666666667,
+      "grad_norm": 0.5545276403427124,
+      "learning_rate": 0.0005,
+      "loss": 0.14,
+      "step": 2600
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.5854529147982063,
+      "eval_loss": 2.8039839267730713,
+      "eval_runtime": 6.5566,
+      "eval_samples_per_second": 76.259,
+      "eval_steps_per_second": 9.609,
+      "step": 2625
+    },
+    {
+      "epoch": 14.0,
+      "eval_exact_match": 14.4,
+      "eval_f1": 22.79698412698414,
+      "eval_qa_bleu": 13.05170441986655,
+      "eval_qa_exact_match": 0.114,
+      "eval_recite_bleu": 17.88705533351519,
+      "eval_recite_exact_match": 0.052,
+      "step": 2625
+    },
+    {
+      "epoch": 14.4,
+      "grad_norm": 0.5276845693588257,
+      "learning_rate": 0.0005,
+      "loss": 0.1305,
+      "step": 2700
+    },
+    {
+      "epoch": 14.933333333333334,
+      "grad_norm": 0.5796266794204712,
+      "learning_rate": 0.0005,
+      "loss": 0.1405,
+      "step": 2800
+    },
+    {
+      "epoch": 14.997333333333334,
+      "eval_accuracy": 0.5847085201793722,
+      "eval_loss": 2.838374137878418,
+      "eval_runtime": 6.2959,
+      "eval_samples_per_second": 79.417,
+      "eval_steps_per_second": 10.007,
+      "step": 2812
+    },
+    {
+      "epoch": 14.997333333333334,
+      "eval_exact_match": 12.8,
+      "eval_f1": 18.866060606060614,
+      "eval_qa_bleu": 11.202036826863646,
+      "eval_qa_exact_match": 0.096,
+      "eval_recite_bleu": 16.939173064657204,
+      "eval_recite_exact_match": 0.03,
+      "step": 2812
+    },
+    {
+      "epoch": 15.466666666666667,
+      "grad_norm": 0.4388948678970337,
+      "learning_rate": 0.0005,
+      "loss": 0.1243,
+      "step": 2900
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 0.5327735543251038,
+      "learning_rate": 0.0005,
+      "loss": 0.1373,
+      "step": 3000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.585067264573991,
+      "eval_loss": 2.83713960647583,
+      "eval_runtime": 6.5308,
+      "eval_samples_per_second": 76.56,
+      "eval_steps_per_second": 9.647,
+      "step": 3000
+    },
+    {
+      "epoch": 16.0,
+      "eval_exact_match": 12.2,
+      "eval_f1": 18.169794584500472,
+      "eval_qa_bleu": 9.673520468570782,
+      "eval_qa_exact_match": 0.092,
+      "eval_recite_bleu": 16.93622051748279,
+      "eval_recite_exact_match": 0.048,
+      "step": 3000
+    },
+    {
+      "epoch": 16.533333333333335,
+      "grad_norm": 0.4463440179824829,
+      "learning_rate": 0.0005,
+      "loss": 0.1192,
+      "step": 3100
+    },
+    {
+      "epoch": 16.997333333333334,
+      "eval_accuracy": 0.5841793721973094,
+      "eval_loss": 2.879470109939575,
+      "eval_runtime": 5.6058,
+      "eval_samples_per_second": 89.194,
+      "eval_steps_per_second": 11.238,
+      "step": 3187
+    },
+    {
+      "epoch": 16.997333333333334,
+      "eval_exact_match": 12.4,
+      "eval_f1": 19.10238095238096,
+      "eval_qa_bleu": 10.318342680515805,
+      "eval_qa_exact_match": 0.098,
+      "eval_recite_bleu": 16.676445435372614,
+      "eval_recite_exact_match": 0.03,
+      "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 sha256:7b6061b8d6135a3f759c6b956a7afa194ea4a0f719fc2e4eb293c4d6c15c2590
+size 5304