Training in progress, epoch 9, checkpoint

checkpoint-4761/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-4761/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ebda6efee3c2f851c7f3d5ecc3c895be7213427a1bab60d92a45436aa538613
+size 143269386

checkpoint-4761/added_tokens.json

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

checkpoint-4761/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56a1f496c76872fad8741862d961aa203078ffaeea7a00bd97269eefd60ddde0
-size 224537202
+oid sha256:5d8c32642e9541e915cdbd2714c3307063b90a99146f89eb79d8536d8cb9070e
+size 286585234

checkpoint-4761/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5266551f6130ade94cf04ccd9552ac46e1b6164e4184ed59bd35096fe21e3b2d
+oid sha256:420c201b3b8c7c108ee6466fd6055fe3d321fb1fcd6036fba7294c27ec521fb5
 size 15024

checkpoint-4761/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:735bed8bd5f21e29ffce33ed2c3923e8c8f6adadd09320467a21d9a2f063570b
+oid sha256:24b6e5e954e9a4d5c4a8d357efa9e0ad30e91a6083c2e86742ae34112a5ecf95
 size 15024

checkpoint-4761/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d6546871770f05e5e151a46fcc5a2cd4d706686f40a546560b9ba8e808ffa4f6
+oid sha256:095482ccd9f4c2f13eb48064407a6d81077f93ca09d0dc3eb8f31d87396e87d0
 size 15024

checkpoint-4761/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2d0863e362cce841c3baa7864c98c610fe995188f6ffcacd6cfbf02425c682d0
+oid sha256:b4d59f9630ca73b11a93ebd10aa490178b77ad2f25c2d308f27bde4dc021549e
 size 15024

checkpoint-4761/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-4761/tokenizer.model

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

checkpoint-4761/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-4761/trainer_state.json

@@ -9,97 +9,87 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1890359168241966,
-      "grad_norm": 0.3393945097923279,
+      "epoch": 0.19,
       "learning_rate": 0.0003,
-      "loss": 1.951,
+      "loss": 1.4758,
       "step": 100
     },
     {
-      "epoch": 0.3780718336483932,
-      "grad_norm": 8017.20361328125,
+      "epoch": 0.38,
       "learning_rate": 0.0003,
-      "loss": 2.0285,
+      "loss": 1.3867,
       "step": 200
     },
     {
-      "epoch": 0.5671077504725898,
-      "grad_norm": 9146.7236328125,
+      "epoch": 0.57,
       "learning_rate": 0.0003,
-      "loss": 6.5404,
+      "loss": 1.4064,
       "step": 300
     },
     {
-      "epoch": 0.7561436672967864,
-      "grad_norm": 574.934326171875,
+      "epoch": 0.76,
       "learning_rate": 0.0003,
-      "loss": 7.7357,
+      "loss": 1.3789,
       "step": 400
     },
     {
-      "epoch": 0.945179584120983,
-      "grad_norm": 25128.55859375,
+      "epoch": 0.95,
       "learning_rate": 0.0003,
-      "loss": 6.3771,
+      "loss": 1.3822,
       "step": 500
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.24974358974358973,
-      "eval_loss": 10.19681453704834,
-      "eval_runtime": 5.1408,
-      "eval_samples_per_second": 97.26,
-      "eval_steps_per_second": 12.255,
+      "eval_accuracy": 0.6171794871794872,
+      "eval_loss": 1.2976911067962646,
+      "eval_runtime": 4.224,
+      "eval_samples_per_second": 118.372,
+      "eval_steps_per_second": 14.915,
       "step": 529
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_exact_match": 0.2,
+      "eval_f1": 2.031878046337681,
       "step": 529
     },
     {
-      "epoch": 1.1342155009451795,
-      "grad_norm": 1332.8885498046875,
+      "epoch": 1.13,
       "learning_rate": 0.0003,
-      "loss": 11.0858,
+      "loss": 1.3048,
       "step": 600
     },
     {
-      "epoch": 1.3232514177693762,
-      "grad_norm": 3666.881591796875,
+      "epoch": 1.32,
       "learning_rate": 0.0003,
-      "loss": 9.483,
+      "loss": 1.2524,
       "step": 700
     },
     {
-      "epoch": 1.5122873345935728,
-      "grad_norm": 356847.0,
+      "epoch": 1.51,
       "learning_rate": 0.0003,
-      "loss": 8.069,
+      "loss": 1.3009,
       "step": 800
     },
     {
-      "epoch": 1.7013232514177694,
-      "grad_norm": 1061.877685546875,
+      "epoch": 1.7,
       "learning_rate": 0.0003,
-      "loss": 8.7493,
+      "loss": 1.268,
       "step": 900
     },
     {
-      "epoch": 1.8903591682419658,
-      "grad_norm": 46670.3046875,
+      "epoch": 1.89,
       "learning_rate": 0.0003,
-      "loss": 8.3415,
+      "loss": 1.2744,
       "step": 1000
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.20507692307692307,
-      "eval_loss": 7.504091262817383,
-      "eval_runtime": 5.2437,
-      "eval_samples_per_second": 95.352,
-      "eval_steps_per_second": 12.014,
+      "eval_accuracy": 0.6032307692307692,
+      "eval_loss": 1.374481201171875,
+      "eval_runtime": 4.4141,
+      "eval_samples_per_second": 113.274,
+      "eval_steps_per_second": 14.272,
       "step": 1058
     },
     {
@@ -109,204 +99,183 @@
       "step": 1058
     },
     {
-      "epoch": 2.0793950850661624,
-      "grad_norm": 2.0727381706237793,
+      "epoch": 2.08,
       "learning_rate": 0.0003,
-      "loss": 8.3511,
+      "loss": 1.2158,
       "step": 1100
     },
     {
-      "epoch": 2.268431001890359,
-      "grad_norm": 1.8185893297195435,
+      "epoch": 2.27,
       "learning_rate": 0.0003,
-      "loss": 7.3065,
+      "loss": 1.1066,
       "step": 1200
     },
     {
-      "epoch": 2.4574669187145557,
-      "grad_norm": 1620.93701171875,
+      "epoch": 2.46,
       "learning_rate": 0.0003,
-      "loss": 7.1302,
+      "loss": 1.1365,
       "step": 1300
     },
     {
-      "epoch": 2.6465028355387523,
-      "grad_norm": 1.0691380500793457,
+      "epoch": 2.65,
       "learning_rate": 0.0003,
-      "loss": 7.4423,
+      "loss": 1.1486,
       "step": 1400
     },
     {
-      "epoch": 2.835538752362949,
-      "grad_norm": 1.0605100393295288,
+      "epoch": 2.84,
       "learning_rate": 0.0003,
-      "loss": 7.6715,
+      "loss": 1.1768,
       "step": 1500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.28507692307692306,
-      "eval_loss": 7.084368705749512,
-      "eval_runtime": 5.493,
-      "eval_samples_per_second": 91.026,
-      "eval_steps_per_second": 11.469,
+      "eval_accuracy": 0.6157435897435898,
+      "eval_loss": 1.331903100013733,
+      "eval_runtime": 4.8466,
+      "eval_samples_per_second": 103.164,
+      "eval_steps_per_second": 12.999,
       "step": 1587
     },
     {
       "epoch": 3.0,
       "eval_exact_match": 0.0,
-      "eval_f1": 0.08,
+      "eval_f1": 0.0,
       "step": 1587
     },
     {
-      "epoch": 3.0245746691871456,
-      "grad_norm": 1967676800.0,
+      "epoch": 3.02,
       "learning_rate": 0.0003,
-      "loss": 7.8617,
+      "loss": 1.1769,
       "step": 1600
     },
     {
-      "epoch": 3.213610586011342,
-      "grad_norm": 83918778793984.0,
+      "epoch": 3.21,
       "learning_rate": 0.0003,
-      "loss": 8.083,
+      "loss": 0.9541,
       "step": 1700
     },
     {
-      "epoch": 3.402646502835539,
-      "grad_norm": 12349449.0,
+      "epoch": 3.4,
       "learning_rate": 0.0003,
-      "loss": 8.0382,
+      "loss": 1.2939,
       "step": 1800
     },
     {
-      "epoch": 3.5916824196597354,
-      "grad_norm": 52939337728.0,
+      "epoch": 3.59,
       "learning_rate": 0.0003,
-      "loss": 7.8761,
+      "loss": 0.9852,
       "step": 1900
     },
     {
-      "epoch": 3.780718336483932,
-      "grad_norm": 1381182.25,
+      "epoch": 3.78,
       "learning_rate": 0.0003,
-      "loss": 7.4665,
+      "loss": 0.9538,
       "step": 2000
     },
     {
-      "epoch": 3.9697542533081287,
-      "grad_norm": 105.80550384521484,
+      "epoch": 3.97,
       "learning_rate": 0.0003,
-      "loss": 7.8182,
+      "loss": 0.9247,
       "step": 2100
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.23933333333333334,
-      "eval_loss": 7.05845308303833,
-      "eval_runtime": 5.0891,
-      "eval_samples_per_second": 98.248,
-      "eval_steps_per_second": 12.379,
+      "eval_accuracy": 0.6101538461538462,
+      "eval_loss": 1.4367107152938843,
+      "eval_runtime": 4.4146,
+      "eval_samples_per_second": 113.259,
+      "eval_steps_per_second": 14.271,
       "step": 2116
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_exact_match": 6.6,
+      "eval_f1": 9.897142857142855,
       "step": 2116
     },
     {
-      "epoch": 4.158790170132325,
-      "grad_norm": 16.110862731933594,
+      "epoch": 4.16,
       "learning_rate": 0.0003,
-      "loss": 7.9728,
+      "loss": 1.1334,
       "step": 2200
     },
     {
-      "epoch": 4.3478260869565215,
-      "grad_norm": 4.362427234649658,
+      "epoch": 4.35,
       "learning_rate": 0.0003,
-      "loss": 7.2875,
+      "loss": 1.5175,
       "step": 2300
     },
     {
-      "epoch": 4.536862003780718,
-      "grad_norm": 18039.095703125,
+      "epoch": 4.54,
       "learning_rate": 0.0003,
-      "loss": 7.3972,
+      "loss": 4.2745,
       "step": 2400
     },
     {
-      "epoch": 4.725897920604915,
-      "grad_norm": 2.3812594051160474e+17,
+      "epoch": 4.73,
       "learning_rate": 0.0003,
-      "loss": 7.4631,
+      "loss": 2.5667,
       "step": 2500
     },
     {
-      "epoch": 4.914933837429111,
-      "grad_norm": 3.9549436569213867,
+      "epoch": 4.91,
       "learning_rate": 0.0003,
-      "loss": 7.3322,
+      "loss": 1.1836,
       "step": 2600
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.22430769230769232,
-      "eval_loss": 6.72421932220459,
-      "eval_runtime": 5.1689,
-      "eval_samples_per_second": 96.731,
-      "eval_steps_per_second": 12.188,
+      "eval_accuracy": 0.556923076923077,
+      "eval_loss": 1.9167817831039429,
+      "eval_runtime": 4.8432,
+      "eval_samples_per_second": 103.238,
+      "eval_steps_per_second": 13.008,
       "step": 2645
     },
     {
       "epoch": 5.0,
       "eval_exact_match": 0.0,
-      "eval_f1": 0.0,
+      "eval_f1": 0.2264729350912329,
       "step": 2645
     },
     {
-      "epoch": 5.103969754253308,
-      "grad_norm": 120114.2421875,
+      "epoch": 5.1,
       "learning_rate": 0.0003,
-      "loss": 7.2428,
+      "loss": 1.7226,
       "step": 2700
     },
     {
-      "epoch": 5.293005671077505,
-      "grad_norm": 391.66119384765625,
+      "epoch": 5.29,
       "learning_rate": 0.0003,
-      "loss": 7.2495,
+      "loss": 2.1466,
       "step": 2800
     },
     {
-      "epoch": 5.482041587901701,
-      "grad_norm": 599371136.0,
+      "epoch": 5.48,
       "learning_rate": 0.0003,
-      "loss": 7.5915,
+      "loss": 1.83,
       "step": 2900
     },
     {
-      "epoch": 5.671077504725898,
-      "grad_norm": 5585.97314453125,
+      "epoch": 5.67,
       "learning_rate": 0.0003,
-      "loss": 7.3014,
+      "loss": 1.8789,
       "step": 3000
     },
     {
-      "epoch": 5.8601134215500945,
-      "grad_norm": 268239548448768.0,
+      "epoch": 5.86,
       "learning_rate": 0.0003,
-      "loss": 7.259,
+      "loss": 2.035,
       "step": 3100
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.24984615384615386,
-      "eval_loss": 6.531904220581055,
-      "eval_runtime": 5.0344,
-      "eval_samples_per_second": 99.317,
-      "eval_steps_per_second": 12.514,
+      "eval_accuracy": 0.5377435897435897,
+      "eval_loss": 2.0794410705566406,
+      "eval_runtime": 4.1284,
+      "eval_samples_per_second": 121.112,
+      "eval_steps_per_second": 15.26,
       "step": 3174
     },
     {
@@ -316,54 +285,48 @@
       "step": 3174
     },
     {
-      "epoch": 6.049149338374291,
-      "grad_norm": 21535.734375,
+      "epoch": 6.05,
       "learning_rate": 0.0003,
-      "loss": 7.5829,
+      "loss": 2.2251,
       "step": 3200
     },
     {
-      "epoch": 6.238185255198488,
-      "grad_norm": 40871530496.0,
+      "epoch": 6.24,
       "learning_rate": 0.0003,
-      "loss": 8.6364,
+      "loss": 2.1457,
       "step": 3300
     },
     {
-      "epoch": 6.427221172022684,
-      "grad_norm": 13.242510795593262,
+      "epoch": 6.43,
       "learning_rate": 0.0003,
-      "loss": 7.6599,
+      "loss": 2.0764,
       "step": 3400
     },
     {
-      "epoch": 6.616257088846881,
-      "grad_norm": 3.0841150283813477,
+      "epoch": 6.62,
       "learning_rate": 0.0003,
-      "loss": 7.5796,
+      "loss": 2.3695,
       "step": 3500
     },
     {
-      "epoch": 6.805293005671078,
-      "grad_norm": 1.6124571561813354,
+      "epoch": 6.81,
       "learning_rate": 0.0003,
-      "loss": 7.3499,
+      "loss": 2.2254,
       "step": 3600
     },
     {
-      "epoch": 6.994328922495274,
-      "grad_norm": 1.8358014822006226,
+      "epoch": 6.99,
       "learning_rate": 0.0003,
-      "loss": 7.2206,
+      "loss": 3.7483,
       "step": 3700
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.22430769230769232,
-      "eval_loss": 6.724715709686279,
-      "eval_runtime": 5.0476,
-      "eval_samples_per_second": 99.056,
-      "eval_steps_per_second": 12.481,
+      "eval_accuracy": 0.48805128205128206,
+      "eval_loss": 2.6722612380981445,
+      "eval_runtime": 4.2095,
+      "eval_samples_per_second": 118.778,
+      "eval_steps_per_second": 14.966,
       "step": 3703
     },
     {
@@ -373,47 +336,42 @@
       "step": 3703
     },
     {
-      "epoch": 7.183364839319471,
-      "grad_norm": 5.786651134490967,
+      "epoch": 7.18,
       "learning_rate": 0.0003,
-      "loss": 7.2316,
+      "loss": 6.5533,
       "step": 3800
     },
     {
-      "epoch": 7.3724007561436675,
-      "grad_norm": 3.7182490825653076,
+      "epoch": 7.37,
       "learning_rate": 0.0003,
-      "loss": 7.2591,
+      "loss": 6.708,
       "step": 3900
     },
     {
-      "epoch": 7.561436672967864,
-      "grad_norm": 31.241104125976562,
+      "epoch": 7.56,
       "learning_rate": 0.0003,
-      "loss": 7.2557,
+      "loss": 7.4301,
       "step": 4000
     },
     {
-      "epoch": 7.750472589792061,
-      "grad_norm": 896688896.0,
+      "epoch": 7.75,
       "learning_rate": 0.0003,
-      "loss": 8.4436,
+      "loss": 8.6731,
       "step": 4100
     },
     {
-      "epoch": 7.939508506616257,
-      "grad_norm": 1.1150538843415904e+19,
+      "epoch": 7.94,
       "learning_rate": 0.0003,
-      "loss": 8.7572,
+      "loss": 7.127,
       "step": 4200
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.479637622833252,
-      "eval_runtime": 5.1473,
-      "eval_samples_per_second": 97.138,
-      "eval_steps_per_second": 12.239,
+      "eval_accuracy": 0.1922051282051282,
+      "eval_loss": 7.040984153747559,
+      "eval_runtime": 4.8262,
+      "eval_samples_per_second": 103.6,
+      "eval_steps_per_second": 13.054,
       "step": 4232
     },
     {
@@ -423,47 +381,42 @@
       "step": 4232
     },
     {
-      "epoch": 8.128544423440454,
-      "grad_norm": 7.2513813972473145,
+      "epoch": 8.13,
       "learning_rate": 0.0003,
-      "loss": 8.1381,
+      "loss": 7.8796,
       "step": 4300
     },
     {
-      "epoch": 8.31758034026465,
-      "grad_norm": 14.863628387451172,
+      "epoch": 8.32,
       "learning_rate": 0.0003,
-      "loss": 7.3335,
+      "loss": 8.0716,
       "step": 4400
     },
     {
-      "epoch": 8.506616257088847,
-      "grad_norm": 5.358551979064941,
+      "epoch": 8.51,
       "learning_rate": 0.0003,
-      "loss": 8.3921,
+      "loss": 8.2395,
       "step": 4500
     },
     {
-      "epoch": 8.695652173913043,
-      "grad_norm": 113.71702575683594,
+      "epoch": 8.7,
       "learning_rate": 0.0003,
-      "loss": 7.221,
+      "loss": 7.9157,
       "step": 4600
     },
     {
-      "epoch": 8.88468809073724,
-      "grad_norm": 2.3997488021850586,
+      "epoch": 8.88,
       "learning_rate": 0.0003,
-      "loss": 8.4835,
+      "loss": 7.5321,
       "step": 4700
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.26184615384615384,
-      "eval_loss": 6.614655017852783,
-      "eval_runtime": 5.026,
-      "eval_samples_per_second": 99.483,
-      "eval_steps_per_second": 12.535,
+      "eval_accuracy": 0.19405128205128205,
+      "eval_loss": 6.64878511428833,
+      "eval_runtime": 4.104,
+      "eval_samples_per_second": 121.831,
+      "eval_steps_per_second": 15.351,
       "step": 4761
     },
     {
@@ -475,23 +428,9 @@
   ],
   "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": 3.875560697500795e+17,
-  "train_batch_size": 2,
+  "total_flos": 8.041718036677263e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-4761/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:186f3ae462881ca0bb3a6b5295a97a245e0a78172be9c14948a2d07a12aa4241
-size 5304
+oid sha256:b59ca157797b57867ac15a1da8ca60d782441a1506bb7cb43703c7748b856900
+size 4728