Training in progress, epoch 7, checkpoint

checkpoint-3703/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-3703/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:158ae91d38b341de8a594d7b53a14ac60a77974209a6e6ba4f94f51e3cb401eb
+size 143269386

checkpoint-3703/added_tokens.json

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

checkpoint-3703/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a97fb3661a0db7baf1cfa3082fd530ebfb9ba1b6882d35c29a0a9f04901739cf
-size 224537202
+oid sha256:9beee29a377964aa512c0ee669c9edda3954a4dcaa1f54b3032a43cee3612e30
+size 286585234

checkpoint-3703/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2978f3e718904406bff8b9da249cc275ca395e058ac4a0580edb9a8bccd21ac1
+oid sha256:88188d8a97480a9ecb61f27b6f2c85687b1a7cf04b086a12b75e69f5ebf625b5
 size 15024

checkpoint-3703/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:670e4466e93414a27126677236c0e5552151a9a31d420652218f2ae1eba07fdf
+oid sha256:75564a9e6969d3cd257d40c41d47500671824b1f8213571e70f11887d0567508
 size 15024

checkpoint-3703/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9caf1d4db9f50bd334919622ccd3a543d622e201e9a8cb415a736ad4da630574
+oid sha256:d4c685cc517dda05a81b44cc7a81fe49d332fafd74430831c2d97fec6f5f6d0f
 size 15024

checkpoint-3703/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8016023a3470a2c15f08808849180c1143d4667399c4838a44e980f95594bb59
+oid sha256:a3bda10f104bbdae64233e0f2bb21c5f8ffbee678ef06a7191eb308ad90790f7
 size 15024

checkpoint-3703/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-3703/tokenizer.model

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

checkpoint-3703/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-3703/trainer_state.json

@@ -9,389 +9,338 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1890359168241966,
-      "grad_norm": 0.5558584332466125,
+      "epoch": 0.19,
       "learning_rate": 3e-05,
-      "loss": 2.1103,
+      "loss": 1.7194,
       "step": 100
     },
     {
-      "epoch": 0.3780718336483932,
-      "grad_norm": 0.4733140468597412,
+      "epoch": 0.38,
       "learning_rate": 3e-05,
-      "loss": 1.931,
+      "loss": 1.4369,
       "step": 200
     },
     {
-      "epoch": 0.5671077504725898,
-      "grad_norm": 0.48580050468444824,
+      "epoch": 0.57,
       "learning_rate": 3e-05,
-      "loss": 1.8948,
+      "loss": 1.4257,
       "step": 300
     },
     {
-      "epoch": 0.7561436672967864,
-      "grad_norm": 0.40655234456062317,
+      "epoch": 0.76,
       "learning_rate": 3e-05,
-      "loss": 1.8988,
+      "loss": 1.3947,
       "step": 400
     },
     {
-      "epoch": 0.945179584120983,
-      "grad_norm": 0.6303955316543579,
+      "epoch": 0.95,
       "learning_rate": 3e-05,
-      "loss": 1.8745,
+      "loss": 1.3948,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5701538461538461,
-      "eval_loss": 1.6455246210098267,
-      "eval_runtime": 5.0805,
-      "eval_samples_per_second": 98.415,
-      "eval_steps_per_second": 12.4,
+      "eval_accuracy": 0.6132307692307692,
+      "eval_loss": 1.3087366819381714,
+      "eval_runtime": 4.3553,
+      "eval_samples_per_second": 114.803,
+      "eval_steps_per_second": 14.465,
       "step": 529
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 15.0,
-      "eval_f1": 23.927619047619068,
+      "eval_exact_match": 24.4,
+      "eval_f1": 37.614420394420414,
       "step": 529
     },
     {
-      "epoch": 1.1342155009451795,
-      "grad_norm": 0.4278818666934967,
+      "epoch": 1.13,
       "learning_rate": 3e-05,
-      "loss": 1.8798,
+      "loss": 1.3927,
       "step": 600
     },
     {
-      "epoch": 1.3232514177693762,
-      "grad_norm": 0.449830025434494,
+      "epoch": 1.32,
       "learning_rate": 3e-05,
-      "loss": 1.8613,
+      "loss": 1.3644,
       "step": 700
     },
     {
-      "epoch": 1.5122873345935728,
-      "grad_norm": 0.5130725502967834,
+      "epoch": 1.51,
       "learning_rate": 3e-05,
-      "loss": 1.8709,
+      "loss": 1.3791,
       "step": 800
     },
     {
-      "epoch": 1.7013232514177694,
-      "grad_norm": 0.4889715611934662,
+      "epoch": 1.7,
       "learning_rate": 3e-05,
-      "loss": 1.8448,
+      "loss": 1.3627,
       "step": 900
     },
     {
-      "epoch": 1.8903591682419658,
-      "grad_norm": 11.186380386352539,
+      "epoch": 1.89,
       "learning_rate": 3e-05,
-      "loss": 1.8764,
+      "loss": 1.3789,
       "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5723589743589743,
-      "eval_loss": 1.6143561601638794,
-      "eval_runtime": 5.0466,
-      "eval_samples_per_second": 99.077,
-      "eval_steps_per_second": 12.484,
+      "eval_accuracy": 0.6146153846153846,
+      "eval_loss": 1.289727807044983,
+      "eval_runtime": 4.4281,
+      "eval_samples_per_second": 112.915,
+      "eval_steps_per_second": 14.227,
       "step": 1058
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 15.6,
-      "eval_f1": 25.825238095238113,
+      "eval_exact_match": 23.2,
+      "eval_f1": 35.56775335775336,
       "step": 1058
     },
     {
-      "epoch": 2.0793950850661624,
-      "grad_norm": 0.5414931774139404,
+      "epoch": 2.08,
       "learning_rate": 3e-05,
-      "loss": 1.8398,
+      "loss": 1.3605,
       "step": 1100
     },
     {
-      "epoch": 2.268431001890359,
-      "grad_norm": 7.441005706787109,
+      "epoch": 2.27,
       "learning_rate": 3e-05,
-      "loss": 1.8348,
+      "loss": 1.3306,
       "step": 1200
     },
     {
-      "epoch": 2.4574669187145557,
-      "grad_norm": 24.312145233154297,
+      "epoch": 2.46,
       "learning_rate": 3e-05,
-      "loss": 1.8421,
+      "loss": 1.3318,
       "step": 1300
     },
     {
-      "epoch": 2.6465028355387523,
-      "grad_norm": 0.7407662868499756,
+      "epoch": 2.65,
       "learning_rate": 3e-05,
-      "loss": 1.8292,
+      "loss": 1.3491,
       "step": 1400
     },
     {
-      "epoch": 2.835538752362949,
-      "grad_norm": 0.9653642177581787,
+      "epoch": 2.84,
       "learning_rate": 3e-05,
-      "loss": 1.828,
+      "loss": 1.3259,
       "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.5734358974358974,
-      "eval_loss": 1.617864727973938,
-      "eval_runtime": 5.2153,
-      "eval_samples_per_second": 95.872,
-      "eval_steps_per_second": 12.08,
+      "eval_accuracy": 0.6178974358974358,
+      "eval_loss": 1.2849316596984863,
+      "eval_runtime": 4.875,
+      "eval_samples_per_second": 102.565,
+      "eval_steps_per_second": 12.923,
       "step": 1587
     },
     {
       "epoch": 3.0,
-      "eval_exact_match": 16.2,
-      "eval_f1": 25.811428571428586,
+      "eval_exact_match": 25.2,
+      "eval_f1": 37.32698412698414,
       "step": 1587
     },
     {
-      "epoch": 3.0245746691871456,
-      "grad_norm": 450.6055603027344,
+      "epoch": 3.02,
       "learning_rate": 3e-05,
-      "loss": 1.8204,
+      "loss": 1.3255,
       "step": 1600
     },
     {
-      "epoch": 3.213610586011342,
-      "grad_norm": 2.1127028465270996,
+      "epoch": 3.21,
       "learning_rate": 3e-05,
-      "loss": 1.8361,
+      "loss": 1.28,
       "step": 1700
     },
     {
-      "epoch": 3.402646502835539,
-      "grad_norm": 564.9320068359375,
+      "epoch": 3.4,
       "learning_rate": 3e-05,
-      "loss": 1.8338,
+      "loss": 1.2995,
       "step": 1800
     },
     {
-      "epoch": 3.5916824196597354,
-      "grad_norm": 52.5078010559082,
+      "epoch": 3.59,
       "learning_rate": 3e-05,
-      "loss": 1.8075,
+      "loss": 1.2991,
       "step": 1900
     },
     {
-      "epoch": 3.780718336483932,
-      "grad_norm": 27.87557601928711,
+      "epoch": 3.78,
       "learning_rate": 3e-05,
-      "loss": 1.8175,
+      "loss": 1.3024,
       "step": 2000
     },
     {
-      "epoch": 3.9697542533081287,
-      "grad_norm": 4.810656547546387,
+      "epoch": 3.97,
       "learning_rate": 3e-05,
-      "loss": 1.8218,
+      "loss": 1.2853,
       "step": 2100
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5728205128205128,
-      "eval_loss": 1.6224406957626343,
-      "eval_runtime": 5.0588,
-      "eval_samples_per_second": 98.838,
-      "eval_steps_per_second": 12.454,
+      "eval_accuracy": 0.615948717948718,
+      "eval_loss": 1.316924810409546,
+      "eval_runtime": 5.3385,
+      "eval_samples_per_second": 93.659,
+      "eval_steps_per_second": 11.801,
       "step": 2116
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 17.4,
-      "eval_f1": 27.194285714285726,
+      "eval_exact_match": 23.8,
+      "eval_f1": 36.284920634920645,
       "step": 2116
     },
     {
-      "epoch": 4.158790170132325,
-      "grad_norm": 13.004076957702637,
+      "epoch": 4.16,
       "learning_rate": 3e-05,
-      "loss": 1.8089,
+      "loss": 1.2575,
       "step": 2200
     },
     {
-      "epoch": 4.3478260869565215,
-      "grad_norm": 2.23368501663208,
+      "epoch": 4.35,
       "learning_rate": 3e-05,
-      "loss": 1.8292,
+      "loss": 1.2261,
       "step": 2300
     },
     {
-      "epoch": 4.536862003780718,
-      "grad_norm": 0.8635586500167847,
+      "epoch": 4.54,
       "learning_rate": 3e-05,
-      "loss": 1.7992,
+      "loss": 1.2536,
       "step": 2400
     },
     {
-      "epoch": 4.725897920604915,
-      "grad_norm": 6.5326738357543945,
+      "epoch": 4.73,
       "learning_rate": 3e-05,
-      "loss": 1.7955,
+      "loss": 1.2376,
       "step": 2500
     },
     {
-      "epoch": 4.914933837429111,
-      "grad_norm": 463.4647521972656,
+      "epoch": 4.91,
       "learning_rate": 3e-05,
-      "loss": 1.7953,
+      "loss": 1.2556,
       "step": 2600
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.5740512820512821,
-      "eval_loss": 1.6217337846755981,
-      "eval_runtime": 5.1898,
-      "eval_samples_per_second": 96.342,
-      "eval_steps_per_second": 12.139,
+      "eval_accuracy": 0.6131794871794872,
+      "eval_loss": 1.3531945943832397,
+      "eval_runtime": 4.4382,
+      "eval_samples_per_second": 112.659,
+      "eval_steps_per_second": 14.195,
       "step": 2645
     },
     {
       "epoch": 5.0,
-      "eval_exact_match": 17.6,
-      "eval_f1": 26.161428571428587,
+      "eval_exact_match": 25.8,
+      "eval_f1": 37.67571428571429,
       "step": 2645
     },
     {
-      "epoch": 5.103969754253308,
-      "grad_norm": 203.82815551757812,
+      "epoch": 5.1,
       "learning_rate": 3e-05,
-      "loss": 1.7845,
+      "loss": 1.2195,
       "step": 2700
     },
     {
-      "epoch": 5.293005671077505,
-      "grad_norm": 660.8076782226562,
+      "epoch": 5.29,
       "learning_rate": 3e-05,
-      "loss": 1.7909,
+      "loss": 1.2003,
       "step": 2800
     },
     {
-      "epoch": 5.482041587901701,
-      "grad_norm": 5727.51806640625,
+      "epoch": 5.48,
       "learning_rate": 3e-05,
-      "loss": 1.7909,
+      "loss": 1.1864,
       "step": 2900
     },
     {
-      "epoch": 5.671077504725898,
-      "grad_norm": 4880698.0,
+      "epoch": 5.67,
       "learning_rate": 3e-05,
-      "loss": 3.5781,
+      "loss": 1.2231,
       "step": 3000
     },
     {
-      "epoch": 5.8601134215500945,
-      "grad_norm": 1192.6517333984375,
+      "epoch": 5.86,
       "learning_rate": 3e-05,
-      "loss": 3.8703,
+      "loss": 1.1972,
       "step": 3100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.5433846153846154,
-      "eval_loss": 1.886108636856079,
-      "eval_runtime": 5.5162,
-      "eval_samples_per_second": 90.643,
-      "eval_steps_per_second": 11.421,
+      "eval_accuracy": 0.6125641025641025,
+      "eval_loss": 1.4135066270828247,
+      "eval_runtime": 5.0312,
+      "eval_samples_per_second": 99.38,
+      "eval_steps_per_second": 12.522,
       "step": 3174
     },
     {
       "epoch": 6.0,
-      "eval_exact_match": 11.4,
-      "eval_f1": 20.358571428571437,
+      "eval_exact_match": 24.0,
+      "eval_f1": 36.140952380952385,
       "step": 3174
     },
     {
-      "epoch": 6.049149338374291,
-      "grad_norm": 532.3778686523438,
+      "epoch": 6.05,
       "learning_rate": 3e-05,
-      "loss": 2.1701,
+      "loss": 1.1889,
       "step": 3200
     },
     {
-      "epoch": 6.238185255198488,
-      "grad_norm": 565.6697998046875,
+      "epoch": 6.24,
       "learning_rate": 3e-05,
-      "loss": 1.8533,
+      "loss": 1.1471,
       "step": 3300
     },
     {
-      "epoch": 6.427221172022684,
-      "grad_norm": 25206.482421875,
+      "epoch": 6.43,
       "learning_rate": 3e-05,
-      "loss": 2.623,
+      "loss": 1.1509,
       "step": 3400
     },
     {
-      "epoch": 6.616257088846881,
-      "grad_norm": 71.88556671142578,
+      "epoch": 6.62,
       "learning_rate": 3e-05,
-      "loss": 3.1573,
+      "loss": 1.1654,
       "step": 3500
     },
     {
-      "epoch": 6.805293005671078,
-      "grad_norm": 269563.15625,
+      "epoch": 6.81,
       "learning_rate": 3e-05,
-      "loss": 2.2696,
+      "loss": 1.1716,
       "step": 3600
     },
     {
-      "epoch": 6.994328922495274,
-      "grad_norm": 11768.7880859375,
+      "epoch": 6.99,
       "learning_rate": 3e-05,
-      "loss": 3.1506,
+      "loss": 1.1839,
       "step": 3700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.2834871794871795,
-      "eval_loss": 8.311681747436523,
-      "eval_runtime": 5.3223,
-      "eval_samples_per_second": 93.944,
-      "eval_steps_per_second": 11.837,
+      "eval_accuracy": 0.6081025641025641,
+      "eval_loss": 1.500714898109436,
+      "eval_runtime": 4.4677,
+      "eval_samples_per_second": 111.916,
+      "eval_steps_per_second": 14.101,
       "step": 3703
     },
     {
       "epoch": 7.0,
-      "eval_exact_match": 0.0,
-      "eval_f1": 0.027458851580988985,
+      "eval_exact_match": 22.8,
+      "eval_f1": 33.44777777777779,
       "step": 3703
     }
   ],
   "logging_steps": 100,
   "max_steps": 26450,
-  "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": 3.012381524235387e+17,
-  "train_batch_size": 2,
+  "total_flos": 6.255076307642614e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-3703/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3895667ea94410ecd283430024ce6f8557816c9d86efd03783de342ef988d79c
-size 5304
+oid sha256:ec9dd2a1e083127818a65cb4d99f38a0a803e54a476491027d2e996644eba47c
+size 4728