Training in progress, step 900, checkpoint

checkpoint-900/adapter_config.json

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": "/kaggle/input/vietcuna-for-binary-classification/kalapa-vietcuna-3b",
+  "base_model_name_or_path": "vietcuna-3b-v2/kalapa-vietcuna-3b/",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -9,10 +9,10 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "lora_alpha": 32,
-  "lora_dropout": 0.1,
+  "lora_dropout": 0.05,
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 8,
+  "r": 16,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [

checkpoint-900/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:228854f44599d5490dc9b9b3d3825441b0b53ef10e8f0db8d15371e93d8fc6ab
-size 9859800
+oid sha256:42b0c7f64f39c1fde49c472891de0b048323463dc53e53ec3da2f0de1f5bb7ce
+size 19690328

checkpoint-900/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7ef97c2a0504f52c6169f9e428e74bb63cb0565ecaad34c8e62044c1170ae43
-size 42788
+oid sha256:d3e36558df938cb7d0bdafa790aa9b44025333a6a0a122a08846ea07212b12d1
+size 38087162

checkpoint-900/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:61ec329a06077f04d198eec7d50c6cf80ae4ba8a8c2ddd9740eb6d3e26087f00
+oid sha256:901c45c1e26d5feb5b174e54aefcb5b0541227a3f1b3bc7bb62d621b3d3909b8
 size 14244

checkpoint-900/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ead4982c5ab6d41af584745fc550a649748926f833c31660855c4ca9a5c74902
+oid sha256:9b74be73c70e57e602c8f5da1b32ba96a19d31c6388dd04946e2c42cd1bab9df
 size 1064

checkpoint-900/special_tokens_map.json

@@ -2,6 +2,6 @@
   "bos_token": "<s>",
   "eos_token": "</s>",
   "pad_token": "<pad>",
-  "sep_token": "<s>",
+  "sep_token": "\n\n",
   "unk_token": "<unk>"
 }

checkpoint-900/tokenizer_config.json

@@ -39,7 +39,7 @@
   "eos_token": "</s>",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "<pad>",
-  "sep_token": "<s>",
+  "sep_token": "\n\n",
   "tokenizer_class": "BloomTokenizer",
   "unk_token": "<unk>"
 }

checkpoint-900/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.5551439523696899,
+  "best_metric": 0.7192752361297607,
   "best_model_checkpoint": "output/checkpoint-400",
-  "epoch": 0.7219492630101273,
+  "epoch": 4.594767070835992,
   "eval_steps": 50,
   "global_step": 900,
   "is_hyper_param_search": false,
@@ -9,287 +9,287 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
-      "learning_rate": 4.444444444444445e-07,
-      "loss": 1.0919,
+      "epoch": 0.01,
+      "learning_rate": 2.777777777777778e-07,
+      "loss": 9.5569,
       "step": 1
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 2e-05,
-      "loss": 0.7661,
+      "epoch": 0.26,
+      "learning_rate": 1.25e-05,
+      "loss": 6.958,
       "step": 50
     },
     {
-      "epoch": 0.04,
-      "eval_accuracy": 0.7142857142857143,
-      "eval_loss": 0.6954500079154968,
-      "eval_runtime": 28.4403,
-      "eval_samples_per_second": 6.153,
-      "eval_steps_per_second": 1.547,
+      "epoch": 0.26,
+      "eval_accuracy": 0.42857142857142855,
+      "eval_loss": 6.581231117248535,
+      "eval_runtime": 28.2966,
+      "eval_samples_per_second": 6.184,
+      "eval_steps_per_second": 1.555,
       "step": 50
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.222222222222222e-05,
-      "loss": 0.6119,
+      "epoch": 0.51,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 1.6141,
       "step": 100
     },
     {
-      "epoch": 0.08,
-      "eval_accuracy": 0.7314285714285714,
-      "eval_loss": 0.5819395780563354,
-      "eval_runtime": 28.3595,
-      "eval_samples_per_second": 6.171,
-      "eval_steps_per_second": 1.552,
+      "epoch": 0.51,
+      "eval_accuracy": 0.5028571428571429,
+      "eval_loss": 1.6640697717666626,
+      "eval_runtime": 28.1139,
+      "eval_samples_per_second": 6.225,
+      "eval_steps_per_second": 1.565,
       "step": 100
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 6.444444444444446e-05,
-      "loss": 0.5869,
+      "epoch": 0.77,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.8533,
       "step": 150
     },
     {
-      "epoch": 0.12,
-      "eval_accuracy": 0.7371428571428571,
-      "eval_loss": 0.6004045605659485,
-      "eval_runtime": 28.3043,
-      "eval_samples_per_second": 6.183,
-      "eval_steps_per_second": 1.555,
+      "epoch": 0.77,
+      "eval_accuracy": 0.6342857142857142,
+      "eval_loss": 1.2631328105926514,
+      "eval_runtime": 26.205,
+      "eval_samples_per_second": 6.678,
+      "eval_steps_per_second": 1.679,
       "step": 150
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.666666666666667e-05,
-      "loss": 0.588,
+      "epoch": 1.02,
+      "learning_rate": 4.908536585365854e-05,
+      "loss": 0.4474,
       "step": 200
     },
     {
-      "epoch": 0.16,
-      "eval_accuracy": 0.76,
-      "eval_loss": 0.5582989454269409,
-      "eval_runtime": 28.3002,
-      "eval_samples_per_second": 6.184,
-      "eval_steps_per_second": 1.555,
+      "epoch": 1.02,
+      "eval_accuracy": 0.7142857142857143,
+      "eval_loss": 0.9036799073219299,
+      "eval_runtime": 26.208,
+      "eval_samples_per_second": 6.677,
+      "eval_steps_per_second": 1.679,
       "step": 200
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 9.80411361410382e-05,
-      "loss": 0.5701,
+      "epoch": 1.28,
+      "learning_rate": 4.603658536585366e-05,
+      "loss": 0.2996,
       "step": 250
     },
     {
-      "epoch": 0.2,
-      "eval_accuracy": 0.7142857142857143,
-      "eval_loss": 0.6456161141395569,
-      "eval_runtime": 28.2527,
-      "eval_samples_per_second": 6.194,
-      "eval_steps_per_second": 1.557,
+      "epoch": 1.28,
+      "eval_accuracy": 0.6742857142857143,
+      "eval_loss": 1.1182247400283813,
+      "eval_runtime": 26.2071,
+      "eval_samples_per_second": 6.678,
+      "eval_steps_per_second": 1.679,
       "step": 250
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 9.31439764936337e-05,
-      "loss": 0.5652,
+      "epoch": 1.53,
+      "learning_rate": 4.298780487804878e-05,
+      "loss": 0.2403,
       "step": 300
     },
     {
-      "epoch": 0.24,
+      "epoch": 1.53,
       "eval_accuracy": 0.72,
-      "eval_loss": 0.6802676916122437,
-      "eval_runtime": 28.3389,
-      "eval_samples_per_second": 6.175,
-      "eval_steps_per_second": 1.553,
+      "eval_loss": 1.076162338256836,
+      "eval_runtime": 26.2142,
+      "eval_samples_per_second": 6.676,
+      "eval_steps_per_second": 1.678,
       "step": 300
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 8.82468168462292e-05,
-      "loss": 0.5936,
+      "epoch": 1.79,
+      "learning_rate": 3.9939024390243905e-05,
+      "loss": 0.1982,
       "step": 350
     },
     {
-      "epoch": 0.28,
-      "eval_accuracy": 0.7542857142857143,
-      "eval_loss": 0.5668751001358032,
-      "eval_runtime": 28.2611,
-      "eval_samples_per_second": 6.192,
-      "eval_steps_per_second": 1.557,
+      "epoch": 1.79,
+      "eval_accuracy": 0.6571428571428571,
+      "eval_loss": 1.4400347471237183,
+      "eval_runtime": 26.2187,
+      "eval_samples_per_second": 6.675,
+      "eval_steps_per_second": 1.678,
       "step": 350
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 8.334965719882468e-05,
-      "loss": 0.5961,
+      "epoch": 2.04,
+      "learning_rate": 3.6890243902439025e-05,
+      "loss": 0.1926,
       "step": 400
     },
     {
-      "epoch": 0.32,
-      "eval_accuracy": 0.76,
-      "eval_loss": 0.5551439523696899,
-      "eval_runtime": 26.2906,
-      "eval_samples_per_second": 6.656,
-      "eval_steps_per_second": 1.674,
+      "epoch": 2.04,
+      "eval_accuracy": 0.7942857142857143,
+      "eval_loss": 0.7192752361297607,
+      "eval_runtime": 26.2201,
+      "eval_samples_per_second": 6.674,
+      "eval_steps_per_second": 1.678,
       "step": 400
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 7.845249755142018e-05,
-      "loss": 0.5842,
+      "epoch": 2.3,
+      "learning_rate": 3.384146341463415e-05,
+      "loss": 0.1258,
       "step": 450
     },
     {
-      "epoch": 0.36,
-      "eval_accuracy": 0.7314285714285714,
-      "eval_loss": 0.6041642427444458,
-      "eval_runtime": 26.2867,
-      "eval_samples_per_second": 6.657,
-      "eval_steps_per_second": 1.674,
+      "epoch": 2.3,
+      "eval_accuracy": 0.7657142857142857,
+      "eval_loss": 1.0054305791854858,
+      "eval_runtime": 26.2093,
+      "eval_samples_per_second": 6.677,
+      "eval_steps_per_second": 1.679,
       "step": 450
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 7.355533790401568e-05,
-      "loss": 0.5749,
+      "epoch": 2.55,
+      "learning_rate": 3.079268292682927e-05,
+      "loss": 0.0857,
       "step": 500
     },
     {
-      "epoch": 0.4,
-      "eval_accuracy": 0.7257142857142858,
-      "eval_loss": 0.6361325979232788,
-      "eval_runtime": 28.3165,
-      "eval_samples_per_second": 6.18,
-      "eval_steps_per_second": 1.554,
+      "epoch": 2.55,
+      "eval_accuracy": 0.7828571428571428,
+      "eval_loss": 1.1228740215301514,
+      "eval_runtime": 26.1929,
+      "eval_samples_per_second": 6.681,
+      "eval_steps_per_second": 1.68,
       "step": 500
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 6.865817825661118e-05,
-      "loss": 0.6174,
+      "epoch": 2.81,
+      "learning_rate": 2.7743902439024393e-05,
+      "loss": 0.1061,
       "step": 550
     },
     {
-      "epoch": 0.44,
-      "eval_accuracy": 0.7485714285714286,
-      "eval_loss": 0.582287073135376,
-      "eval_runtime": 28.3412,
-      "eval_samples_per_second": 6.175,
-      "eval_steps_per_second": 1.553,
+      "epoch": 2.81,
+      "eval_accuracy": 0.7657142857142857,
+      "eval_loss": 1.2351529598236084,
+      "eval_runtime": 26.2161,
+      "eval_samples_per_second": 6.675,
+      "eval_steps_per_second": 1.678,
       "step": 550
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 6.376101860920667e-05,
-      "loss": 0.5752,
+      "epoch": 3.06,
+      "learning_rate": 2.4695121951219512e-05,
+      "loss": 0.103,
       "step": 600
     },
     {
-      "epoch": 0.48,
-      "eval_accuracy": 0.7314285714285714,
-      "eval_loss": 0.6466975212097168,
-      "eval_runtime": 28.3593,
-      "eval_samples_per_second": 6.171,
-      "eval_steps_per_second": 1.552,
+      "epoch": 3.06,
+      "eval_accuracy": 0.7542857142857143,
+      "eval_loss": 1.413483738899231,
+      "eval_runtime": 26.2198,
+      "eval_samples_per_second": 6.674,
+      "eval_steps_per_second": 1.678,
       "step": 600
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 5.886385896180215e-05,
-      "loss": 0.5679,
+      "epoch": 3.32,
+      "learning_rate": 2.1646341463414634e-05,
+      "loss": 0.0588,
       "step": 650
     },
     {
-      "epoch": 0.52,
-      "eval_accuracy": 0.72,
-      "eval_loss": 0.5773769617080688,
-      "eval_runtime": 28.2118,
-      "eval_samples_per_second": 6.203,
-      "eval_steps_per_second": 1.56,
+      "epoch": 3.32,
+      "eval_accuracy": 0.7542857142857143,
+      "eval_loss": 1.685744285583496,
+      "eval_runtime": 26.2164,
+      "eval_samples_per_second": 6.675,
+      "eval_steps_per_second": 1.678,
       "step": 650
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 5.396669931439765e-05,
-      "loss": 0.5793,
+      "epoch": 3.57,
+      "learning_rate": 1.8597560975609757e-05,
+      "loss": 0.0871,
       "step": 700
     },
     {
-      "epoch": 0.56,
-      "eval_accuracy": 0.6914285714285714,
-      "eval_loss": 0.5711085796356201,
-      "eval_runtime": 28.2951,
-      "eval_samples_per_second": 6.185,
-      "eval_steps_per_second": 1.555,
+      "epoch": 3.57,
+      "eval_accuracy": 0.7885714285714286,
+      "eval_loss": 1.5099748373031616,
+      "eval_runtime": 26.2038,
+      "eval_samples_per_second": 6.678,
+      "eval_steps_per_second": 1.679,
       "step": 700
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 4.906953966699315e-05,
-      "loss": 0.5897,
+      "epoch": 3.83,
+      "learning_rate": 1.554878048780488e-05,
+      "loss": 0.0539,
       "step": 750
     },
     {
-      "epoch": 0.6,
-      "eval_accuracy": 0.7028571428571428,
-      "eval_loss": 0.5825155377388,
-      "eval_runtime": 28.19,
-      "eval_samples_per_second": 6.208,
-      "eval_steps_per_second": 1.561,
+      "epoch": 3.83,
+      "eval_accuracy": 0.7714285714285715,
+      "eval_loss": 1.7970468997955322,
+      "eval_runtime": 26.2034,
+      "eval_samples_per_second": 6.679,
+      "eval_steps_per_second": 1.679,
       "step": 750
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 4.4172380019588636e-05,
-      "loss": 0.5782,
+      "epoch": 4.08,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0339,
       "step": 800
     },
     {
-      "epoch": 0.64,
-      "eval_accuracy": 0.7257142857142858,
-      "eval_loss": 0.5976516008377075,
-      "eval_runtime": 28.2521,
-      "eval_samples_per_second": 6.194,
-      "eval_steps_per_second": 1.557,
+      "epoch": 4.08,
+      "eval_accuracy": 0.7714285714285715,
+      "eval_loss": 1.7582976818084717,
+      "eval_runtime": 26.2084,
+      "eval_samples_per_second": 6.677,
+      "eval_steps_per_second": 1.679,
       "step": 800
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 3.927522037218414e-05,
-      "loss": 0.593,
+      "epoch": 4.34,
+      "learning_rate": 9.451219512195123e-06,
+      "loss": 0.0279,
       "step": 850
     },
     {
-      "epoch": 0.68,
-      "eval_accuracy": 0.7257142857142858,
-      "eval_loss": 0.6304438710212708,
-      "eval_runtime": 28.304,
-      "eval_samples_per_second": 6.183,
-      "eval_steps_per_second": 1.555,
+      "epoch": 4.34,
+      "eval_accuracy": 0.7714285714285715,
+      "eval_loss": 1.7328156232833862,
+      "eval_runtime": 26.2197,
+      "eval_samples_per_second": 6.674,
+      "eval_steps_per_second": 1.678,
       "step": 850
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 3.437806072477963e-05,
-      "loss": 0.5624,
+      "epoch": 4.59,
+      "learning_rate": 6.402439024390244e-06,
+      "loss": 0.0271,
       "step": 900
     },
     {
-      "epoch": 0.72,
-      "eval_accuracy": 0.72,
-      "eval_loss": 0.5925478935241699,
-      "eval_runtime": 27.995,
-      "eval_samples_per_second": 6.251,
-      "eval_steps_per_second": 1.572,
+      "epoch": 4.59,
+      "eval_accuracy": 0.76,
+      "eval_loss": 2.002990484237671,
+      "eval_runtime": 26.219,
+      "eval_samples_per_second": 6.675,
+      "eval_steps_per_second": 1.678,
       "step": 900
     }
   ],
   "logging_steps": 50,
-  "max_steps": 1246,
-  "num_train_epochs": 1,
+  "max_steps": 1000,
+  "num_train_epochs": 6,
   "save_steps": 50,
-  "total_flos": 3.3378359454142464e+17,
+  "total_flos": 3.356316829903872e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-900/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b25ecbcf37659c2f593e1603dab54c3eaea43862c7f192a7e6df4c4dcc86d20d
+oid sha256:5a0ef36d7c31265f68a3edc8ae53ce9272dcc90a79f261e29a2a46e040c77a33
 size 4600