Training in progress, epoch 8, checkpoint

checkpoint-3073/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-3073/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca3eba3e97cea3eca1137120d627d6731a8c33858a3fb1bdf2f952a058cb5813
+size 143269386

checkpoint-3073/added_tokens.json

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

checkpoint-3073/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b46e7fd57ce278b74cae128871812a7d6598e5e51ecd1b8278d00d3a3bf397cd
-size 224537202
+oid sha256:f86bed739353a8429553b53d4d569f780fe45bb5f73ec3158f2582784ee48f28
+size 286585234

checkpoint-3073/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a8f60b350eb068b8df170a9b0555f3135e6ab0a09369e3d16f4bade064529d16
+oid sha256:300b860a9864f6175dcf9450fca618a256f504b66efa47f93e67ae3334efe4b9
 size 15024

checkpoint-3073/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e735a671ec875e5ef4adcb207df2e1a3df9ea947bc180d7082f3f7db98fc270d
+oid sha256:18ea9895ee88c070b8f60aa6d04095c2fe316be0e8fc98961570431c3286af0c
 size 15024

checkpoint-3073/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b145b574a19314e6f08a02f4852d6ef83feed75f0d89e3aff2d6fa616294e70
+oid sha256:c103d1ae7cf77e9b8b3e2f9f7f50829dd6aba3aee9fb9269043aa7319ed96d51
 size 15024

checkpoint-3073/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aba9a26fbb6e0d8ecacb12208ee15249b962d7133e641f0bd91bf288d0873f9b
+oid sha256:0b51b9a206b7af5b0255f08c7088e20f982857f0b0c3b504a41a795374a6f020
 size 15024

checkpoint-3073/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-3073/tokenizer.model

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

checkpoint-3073/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-3073/trainer_state.json

@@ -9,379 +9,326 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.29282576866764276,
-      "grad_norm": 0.2411503791809082,
+      "epoch": 0.29,
       "learning_rate": 0.0001,
-      "loss": 2.0009,
+      "loss": 1.4842,
       "step": 100
     },
     {
-      "epoch": 0.5856515373352855,
-      "grad_norm": 0.22286373376846313,
+      "epoch": 0.59,
       "learning_rate": 0.0001,
-      "loss": 1.9538,
+      "loss": 1.4016,
       "step": 200
     },
     {
-      "epoch": 0.8784773060029283,
-      "grad_norm": 0.24079065024852753,
+      "epoch": 0.88,
       "learning_rate": 0.0001,
-      "loss": 1.9571,
+      "loss": 1.3892,
       "step": 300
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_accuracy": 0.45379487179487177,
-      "eval_loss": 3.9512319564819336,
-      "eval_runtime": 5.0647,
-      "eval_samples_per_second": 98.722,
-      "eval_steps_per_second": 12.439,
+      "epoch": 1.0,
+      "eval_accuracy": 0.4543589743589744,
+      "eval_loss": 3.4055869579315186,
+      "eval_runtime": 4.1688,
+      "eval_samples_per_second": 119.938,
+      "eval_steps_per_second": 15.112,
       "step": 341
     },
     {
-      "epoch": 0.9985358711566618,
-      "eval_bleu": 0.28767194245703276,
-      "eval_exact_match": 0.03,
-      "eval_prefix_exact_match": 0.042,
+      "epoch": 1.0,
+      "eval_bleu": 0.09711501224106496,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.054,
       "step": 341
     },
     {
-      "epoch": 1.171303074670571,
-      "grad_norm": 0.33546799421310425,
+      "epoch": 1.17,
       "learning_rate": 0.0001,
-      "loss": 1.9255,
+      "loss": 1.3608,
       "step": 400
     },
     {
-      "epoch": 1.4641288433382138,
-      "grad_norm": 0.3515666723251343,
+      "epoch": 1.46,
       "learning_rate": 0.0001,
-      "loss": 1.8684,
+      "loss": 1.3456,
       "step": 500
     },
     {
-      "epoch": 1.7569546120058566,
-      "grad_norm": 0.35733896493911743,
+      "epoch": 1.76,
       "learning_rate": 0.0001,
-      "loss": 1.8819,
+      "loss": 1.3499,
       "step": 600
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.4482564102564103,
-      "eval_loss": 4.112804412841797,
-      "eval_runtime": 5.0848,
-      "eval_samples_per_second": 98.333,
-      "eval_steps_per_second": 12.39,
+      "eval_accuracy": 0.4576923076923077,
+      "eval_loss": 3.453113317489624,
+      "eval_runtime": 4.4031,
+      "eval_samples_per_second": 113.556,
+      "eval_steps_per_second": 14.308,
       "step": 683
     },
     {
       "epoch": 2.0,
-      "eval_bleu": 0.24521038067633488,
-      "eval_exact_match": 0.02,
-      "eval_prefix_exact_match": 0.044,
+      "eval_bleu": 0.1225593574040313,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 683
     },
     {
-      "epoch": 2.049780380673499,
-      "grad_norm": 0.3784458339214325,
+      "epoch": 2.05,
       "learning_rate": 0.0001,
-      "loss": 1.8613,
+      "loss": 1.3238,
       "step": 700
     },
     {
-      "epoch": 2.342606149341142,
-      "grad_norm": 0.46702849864959717,
+      "epoch": 2.34,
       "learning_rate": 0.0001,
-      "loss": 1.743,
+      "loss": 1.2486,
       "step": 800
     },
     {
-      "epoch": 2.6354319180087846,
-      "grad_norm": 0.5156407356262207,
+      "epoch": 2.64,
       "learning_rate": 0.0001,
-      "loss": 1.7653,
+      "loss": 1.2452,
       "step": 900
     },
     {
-      "epoch": 2.9282576866764276,
-      "grad_norm": 0.4755038619041443,
+      "epoch": 2.93,
       "learning_rate": 0.0001,
-      "loss": 1.7702,
+      "loss": 1.2427,
       "step": 1000
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_accuracy": 0.446051282051282,
-      "eval_loss": 4.327652454376221,
-      "eval_runtime": 5.1641,
-      "eval_samples_per_second": 96.822,
-      "eval_steps_per_second": 12.2,
+      "epoch": 3.0,
+      "eval_accuracy": 0.4584102564102564,
+      "eval_loss": 3.671137809753418,
+      "eval_runtime": 4.4094,
+      "eval_samples_per_second": 113.393,
+      "eval_steps_per_second": 14.288,
       "step": 1024
     },
     {
-      "epoch": 2.998535871156662,
-      "eval_bleu": 0.2598440457703149,
-      "eval_exact_match": 0.018,
-      "eval_prefix_exact_match": 0.034,
+      "epoch": 3.0,
+      "eval_bleu": 0.13165889774889974,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.056,
       "step": 1024
     },
     {
-      "epoch": 3.22108345534407,
-      "grad_norm": 0.5288150906562805,
+      "epoch": 3.22,
       "learning_rate": 0.0001,
-      "loss": 1.6325,
+      "loss": 1.1507,
       "step": 1100
     },
     {
-      "epoch": 3.513909224011713,
-      "grad_norm": 0.5784010887145996,
+      "epoch": 3.51,
       "learning_rate": 0.0001,
-      "loss": 1.5896,
+      "loss": 1.1218,
       "step": 1200
     },
     {
-      "epoch": 3.8067349926793557,
-      "grad_norm": 0.6458429098129272,
+      "epoch": 3.81,
       "learning_rate": 0.0001,
-      "loss": 1.6163,
+      "loss": 1.1231,
       "step": 1300
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.44235897435897437,
-      "eval_loss": 4.584905624389648,
-      "eval_runtime": 4.9243,
-      "eval_samples_per_second": 101.538,
-      "eval_steps_per_second": 12.794,
+      "eval_accuracy": 0.45697435897435895,
+      "eval_loss": 3.7999510765075684,
+      "eval_runtime": 4.0886,
+      "eval_samples_per_second": 122.29,
+      "eval_steps_per_second": 15.409,
       "step": 1366
     },
     {
       "epoch": 4.0,
-      "eval_bleu": 0.27967222841565076,
-      "eval_exact_match": 0.01,
-      "eval_prefix_exact_match": 0.038,
+      "eval_bleu": 0.15413978034957934,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.054,
       "step": 1366
     },
     {
-      "epoch": 4.099560761346998,
-      "grad_norm": 0.732841432094574,
+      "epoch": 4.1,
       "learning_rate": 0.0001,
-      "loss": 1.5441,
+      "loss": 1.0823,
       "step": 1400
     },
     {
-      "epoch": 4.392386530014641,
-      "grad_norm": 0.7372816205024719,
+      "epoch": 4.39,
       "learning_rate": 0.0001,
-      "loss": 1.4142,
+      "loss": 0.997,
       "step": 1500
     },
     {
-      "epoch": 4.685212298682284,
-      "grad_norm": 0.8235049247741699,
+      "epoch": 4.69,
       "learning_rate": 0.0001,
-      "loss": 1.4198,
+      "loss": 1.0024,
       "step": 1600
     },
     {
-      "epoch": 4.978038067349927,
-      "grad_norm": 0.7765743732452393,
+      "epoch": 4.98,
       "learning_rate": 0.0001,
-      "loss": 1.4427,
+      "loss": 0.995,
       "step": 1700
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_accuracy": 0.43861538461538463,
-      "eval_loss": 4.850330829620361,
-      "eval_runtime": 4.9395,
-      "eval_samples_per_second": 101.224,
-      "eval_steps_per_second": 12.754,
+      "epoch": 5.0,
+      "eval_accuracy": 0.4552307692307692,
+      "eval_loss": 3.953216314315796,
+      "eval_runtime": 4.9824,
+      "eval_samples_per_second": 100.354,
+      "eval_steps_per_second": 12.645,
       "step": 1707
     },
     {
-      "epoch": 4.998535871156662,
-      "eval_bleu": 0.2170008640509292,
-      "eval_exact_match": 0.006,
-      "eval_prefix_exact_match": 0.032,
+      "epoch": 5.0,
+      "eval_bleu": 0.1611344512465791,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.046,
       "step": 1707
     },
     {
-      "epoch": 5.270863836017569,
-      "grad_norm": 0.9119400978088379,
+      "epoch": 5.27,
       "learning_rate": 0.0001,
-      "loss": 1.2469,
+      "loss": 0.8495,
       "step": 1800
     },
     {
-      "epoch": 5.563689604685212,
-      "grad_norm": 0.8966185450553894,
+      "epoch": 5.56,
       "learning_rate": 0.0001,
-      "loss": 1.2294,
+      "loss": 0.854,
       "step": 1900
     },
     {
-      "epoch": 5.856515373352855,
-      "grad_norm": 0.9629669189453125,
+      "epoch": 5.86,
       "learning_rate": 0.0001,
-      "loss": 1.2498,
+      "loss": 0.8693,
       "step": 2000
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.4348717948717949,
-      "eval_loss": 5.09260892868042,
-      "eval_runtime": 4.9572,
-      "eval_samples_per_second": 100.863,
-      "eval_steps_per_second": 12.709,
+      "eval_accuracy": 0.45261538461538464,
+      "eval_loss": 4.0766072273254395,
+      "eval_runtime": 4.8095,
+      "eval_samples_per_second": 103.961,
+      "eval_steps_per_second": 13.099,
       "step": 2049
     },
     {
       "epoch": 6.0,
-      "eval_bleu": 0.20196479624851718,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.03,
+      "eval_bleu": 0.1489904300924264,
+      "eval_exact_match": 0.0,
+      "eval_prefix_exact_match": 0.044,
       "step": 2049
     },
     {
-      "epoch": 6.149341142020498,
-      "grad_norm": 1.0808535814285278,
+      "epoch": 6.15,
       "learning_rate": 0.0001,
-      "loss": 1.1238,
+      "loss": 0.7917,
       "step": 2100
     },
     {
-      "epoch": 6.44216691068814,
-      "grad_norm": 1.1639196872711182,
+      "epoch": 6.44,
       "learning_rate": 0.0001,
-      "loss": 1.0301,
+      "loss": 0.7061,
       "step": 2200
     },
     {
-      "epoch": 6.734992679355783,
-      "grad_norm": 1.0662291049957275,
+      "epoch": 6.73,
       "learning_rate": 0.0001,
-      "loss": 1.0655,
+      "loss": 0.7302,
       "step": 2300
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_accuracy": 0.4326153846153846,
-      "eval_loss": 5.270845890045166,
-      "eval_runtime": 5.089,
-      "eval_samples_per_second": 98.251,
-      "eval_steps_per_second": 12.38,
+      "epoch": 7.0,
+      "eval_accuracy": 0.4501025641025641,
+      "eval_loss": 4.171727180480957,
+      "eval_runtime": 4.2169,
+      "eval_samples_per_second": 118.57,
+      "eval_steps_per_second": 14.94,
       "step": 2390
     },
     {
-      "epoch": 6.998535871156662,
-      "eval_bleu": 0.17780932376070374,
-      "eval_exact_match": 0.004,
-      "eval_prefix_exact_match": 0.024,
+      "epoch": 7.0,
       "step": 2390
     },
     {
-      "epoch": 7.027818448023426,
-      "grad_norm": 1.6462695598602295,
+      "epoch": 7.03,
       "learning_rate": 0.0001,
-      "loss": 1.0358,
+      "loss": 0.7227,
       "step": 2400
     },
     {
-      "epoch": 7.320644216691068,
-      "grad_norm": 1.2542779445648193,
+      "epoch": 7.32,
       "learning_rate": 0.0001,
-      "loss": 0.8413,
+      "loss": 0.573,
       "step": 2500
     },
     {
-      "epoch": 7.613469985358711,
-      "grad_norm": 1.334762454032898,
+      "epoch": 7.61,
       "learning_rate": 0.0001,
-      "loss": 0.8654,
+      "loss": 0.6036,
       "step": 2600
     },
     {
-      "epoch": 7.906295754026354,
-      "grad_norm": 1.2401787042617798,
+      "epoch": 7.91,
       "learning_rate": 0.0001,
-      "loss": 0.8733,
+      "loss": 0.6033,
       "step": 2700
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.4316923076923077,
-      "eval_loss": 5.402440547943115,
-      "eval_runtime": 4.9383,
-      "eval_samples_per_second": 101.25,
-      "eval_steps_per_second": 12.758,
+      "eval_accuracy": 0.448,
+      "eval_loss": 4.277780055999756,
+      "eval_runtime": 4.0947,
+      "eval_samples_per_second": 122.11,
+      "eval_steps_per_second": 15.386,
       "step": 2732
     },
     {
       "epoch": 8.0,
-      "eval_bleu": 0.21418356891762502,
-      "eval_exact_match": 0.002,
-      "eval_prefix_exact_match": 0.026,
       "step": 2732
     },
     {
-      "epoch": 8.199121522693996,
-      "grad_norm": 1.255996584892273,
+      "epoch": 8.2,
       "learning_rate": 0.0001,
-      "loss": 0.7302,
+      "loss": 0.4945,
       "step": 2800
     },
     {
-      "epoch": 8.49194729136164,
-      "grad_norm": 1.2836525440216064,
+      "epoch": 8.49,
       "learning_rate": 0.0001,
-      "loss": 0.6875,
+      "loss": 0.4718,
       "step": 2900
     },
     {
-      "epoch": 8.784773060029282,
-      "grad_norm": 1.375715732574463,
+      "epoch": 8.78,
       "learning_rate": 0.0001,
-      "loss": 0.7219,
+      "loss": 0.4825,
       "step": 3000
     },
     {
-      "epoch": 8.998535871156662,
-      "eval_accuracy": 0.42938461538461536,
-      "eval_loss": 5.534832000732422,
-      "eval_runtime": 5.2363,
-      "eval_samples_per_second": 95.487,
-      "eval_steps_per_second": 12.031,
+      "epoch": 9.0,
+      "eval_accuracy": 0.4462051282051282,
+      "eval_loss": 4.34151554107666,
+      "eval_runtime": 4.8113,
+      "eval_samples_per_second": 103.921,
+      "eval_steps_per_second": 13.094,
       "step": 3073
     },
     {
-      "epoch": 8.998535871156662,
-      "eval_bleu": 0.20459866130314297,
-      "eval_exact_match": 0.0,
-      "eval_prefix_exact_match": 0.024,
+      "epoch": 9.0,
       "step": 3073
     }
   ],
   "logging_steps": 100,
   "max_steps": 17050,
-  "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": 2.639223295311872e+17,
-  "train_batch_size": 1,
+  "total_flos": 6.087152951372022e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-3073/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7e63e74298c414f7364874e60281a380506bddfac466373e2426a740b4eb5594
-size 5304
+oid sha256:3c60050af374bb2df141f4673c8141d59d83a5bd892d35ffa56ce8af555309ff
+size 4728