Training in progress, step 1050, checkpoint

checkpoint-1050/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-1050/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84d174e10963e61e616b25042f10241f7091f8d0dfec7d1b8162c6b8485053ef
-size 9859800
+oid sha256:ed3c9abdcaebd8fed90b0e9dbcfc58d338cddac8c98ed09e9eb3c2a1dc453474
+size 19690328

checkpoint-1050/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:29a08c8050fd7d2bcc8a9beffcd833c45f5ef7cba2ae16f80991bd1ff36bd790
-size 42788
+oid sha256:3eea3f7b948c2b33853f249d3e69972f34610f33597cda8caff6af703e714d43
+size 38087162

checkpoint-1050/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2889f49651bbfe7948eb19c06b05d1448af5834c552da2e8af763533e42ba477
+oid sha256:cbc4ff4a27c9256c02f945a8115b875fa7a23ad731151aaf932eef63b5f491be
 size 14244

checkpoint-1050/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:03f52f2d80572b65aeef414e050b0f94719a8b87cfec7ca18761fd55a4356282
+oid sha256:8df07083304f589cfb4fe8c46d90a12aca106e53336b883c0155d2a42358d55f
 size 1064

checkpoint-1050/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-1050/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-1050/trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.5551439523696899,
-  "best_model_checkpoint": "output/checkpoint-400",
+  "best_metric": 0.45803794264793396,
+  "best_model_checkpoint": "output/checkpoint-150",
   "epoch": 0.8422741401784819,
   "eval_steps": 50,
   "global_step": 1050,
@@ -10,323 +10,323 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 4.444444444444445e-07,
-      "loss": 1.0919,
+      "learning_rate": 0.0,
+      "loss": 1.6453,
       "step": 1
     },
     {
       "epoch": 0.04,
-      "learning_rate": 2e-05,
-      "loss": 0.7661,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 1.3216,
       "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,
+      "eval_accuracy": 0.8057142857142857,
+      "eval_loss": 0.6264824271202087,
+      "eval_runtime": 28.8159,
+      "eval_samples_per_second": 6.073,
+      "eval_steps_per_second": 1.527,
       "step": 50
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.222222222222222e-05,
-      "loss": 0.6119,
+      "learning_rate": 2.088888888888889e-05,
+      "loss": 0.918,
       "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,
+      "eval_accuracy": 0.7885714285714286,
+      "eval_loss": 0.49068325757980347,
+      "eval_runtime": 26.1642,
+      "eval_samples_per_second": 6.689,
+      "eval_steps_per_second": 1.682,
       "step": 100
     },
     {
       "epoch": 0.12,
-      "learning_rate": 6.444444444444446e-05,
-      "loss": 0.5869,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.7219,
       "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,
+      "eval_accuracy": 0.7942857142857143,
+      "eval_loss": 0.45803794264793396,
+      "eval_runtime": 26.1308,
+      "eval_samples_per_second": 6.697,
+      "eval_steps_per_second": 1.684,
       "step": 150
     },
     {
       "epoch": 0.16,
-      "learning_rate": 8.666666666666667e-05,
-      "loss": 0.588,
+      "learning_rate": 4.311111111111111e-05,
+      "loss": 0.6619,
       "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,
+      "eval_accuracy": 0.8285714285714286,
+      "eval_loss": 0.4676758646965027,
+      "eval_runtime": 26.1149,
+      "eval_samples_per_second": 6.701,
+      "eval_steps_per_second": 1.685,
       "step": 200
     },
     {
       "epoch": 0.2,
-      "learning_rate": 9.80411361410382e-05,
-      "loss": 0.5701,
+      "learning_rate": 4.906953966699315e-05,
+      "loss": 0.633,
       "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,
+      "eval_accuracy": 0.8114285714285714,
+      "eval_loss": 0.4800652265548706,
+      "eval_runtime": 26.1161,
+      "eval_samples_per_second": 6.701,
+      "eval_steps_per_second": 1.685,
       "step": 250
     },
     {
       "epoch": 0.24,
-      "learning_rate": 9.31439764936337e-05,
-      "loss": 0.5652,
+      "learning_rate": 4.66209598432909e-05,
+      "loss": 0.6147,
       "step": 300
     },
     {
       "epoch": 0.24,
-      "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_accuracy": 0.7657142857142857,
+      "eval_loss": 0.5112562775611877,
+      "eval_runtime": 26.1127,
+      "eval_samples_per_second": 6.702,
+      "eval_steps_per_second": 1.685,
       "step": 300
     },
     {
       "epoch": 0.28,
-      "learning_rate": 8.82468168462292e-05,
-      "loss": 0.5936,
+      "learning_rate": 4.4172380019588636e-05,
+      "loss": 0.6355,
       "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,
+      "eval_accuracy": 0.7828571428571428,
+      "eval_loss": 0.5210281610488892,
+      "eval_runtime": 26.1048,
+      "eval_samples_per_second": 6.704,
+      "eval_steps_per_second": 1.686,
       "step": 350
     },
     {
       "epoch": 0.32,
-      "learning_rate": 8.334965719882468e-05,
-      "loss": 0.5961,
+      "learning_rate": 4.1723800195886386e-05,
+      "loss": 0.593,
       "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,
+      "eval_accuracy": 0.7885714285714286,
+      "eval_loss": 0.5098910927772522,
+      "eval_runtime": 26.1072,
+      "eval_samples_per_second": 6.703,
+      "eval_steps_per_second": 1.685,
       "step": 400
     },
     {
       "epoch": 0.36,
-      "learning_rate": 7.845249755142018e-05,
-      "loss": 0.5842,
+      "learning_rate": 3.927522037218414e-05,
+      "loss": 0.6047,
       "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,
+      "eval_loss": 0.5602952837944031,
+      "eval_runtime": 26.1116,
+      "eval_samples_per_second": 6.702,
+      "eval_steps_per_second": 1.685,
       "step": 450
     },
     {
       "epoch": 0.4,
-      "learning_rate": 7.355533790401568e-05,
-      "loss": 0.5749,
+      "learning_rate": 3.682664054848188e-05,
+      "loss": 0.5661,
       "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,
+      "eval_accuracy": 0.7542857142857143,
+      "eval_loss": 0.5670996904373169,
+      "eval_runtime": 26.1104,
+      "eval_samples_per_second": 6.702,
+      "eval_steps_per_second": 1.685,
       "step": 500
     },
     {
       "epoch": 0.44,
-      "learning_rate": 6.865817825661118e-05,
-      "loss": 0.6174,
+      "learning_rate": 3.437806072477963e-05,
+      "loss": 0.6111,
       "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,
+      "eval_accuracy": 0.7714285714285715,
+      "eval_loss": 0.5074692368507385,
+      "eval_runtime": 26.1037,
+      "eval_samples_per_second": 6.704,
+      "eval_steps_per_second": 1.686,
       "step": 550
     },
     {
       "epoch": 0.48,
-      "learning_rate": 6.376101860920667e-05,
-      "loss": 0.5752,
+      "learning_rate": 3.1929480901077375e-05,
+      "loss": 0.5788,
       "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,
+      "eval_accuracy": 0.7485714285714286,
+      "eval_loss": 0.6720603108406067,
+      "eval_runtime": 26.1036,
+      "eval_samples_per_second": 6.704,
+      "eval_steps_per_second": 1.686,
       "step": 600
     },
     {
       "epoch": 0.52,
-      "learning_rate": 5.886385896180215e-05,
-      "loss": 0.5679,
+      "learning_rate": 2.9480901077375122e-05,
+      "loss": 0.562,
       "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,
+      "eval_accuracy": 0.7942857142857143,
+      "eval_loss": 0.5063876509666443,
+      "eval_runtime": 26.1086,
+      "eval_samples_per_second": 6.703,
+      "eval_steps_per_second": 1.685,
       "step": 650
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5.396669931439765e-05,
-      "loss": 0.5793,
+      "learning_rate": 2.7032321253672872e-05,
+      "loss": 0.5472,
       "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,
+      "eval_accuracy": 0.7371428571428571,
+      "eval_loss": 0.5650284886360168,
+      "eval_runtime": 26.1065,
+      "eval_samples_per_second": 6.703,
+      "eval_steps_per_second": 1.685,
       "step": 700
     },
     {
       "epoch": 0.6,
-      "learning_rate": 4.906953966699315e-05,
-      "loss": 0.5897,
+      "learning_rate": 2.458374142997062e-05,
+      "loss": 0.551,
       "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,
+      "eval_accuracy": 0.8,
+      "eval_loss": 0.5052895545959473,
+      "eval_runtime": 26.1094,
+      "eval_samples_per_second": 6.703,
+      "eval_steps_per_second": 1.685,
       "step": 750
     },
     {
       "epoch": 0.64,
-      "learning_rate": 4.4172380019588636e-05,
-      "loss": 0.5782,
+      "learning_rate": 2.2135161606268366e-05,
+      "loss": 0.5259,
       "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,
+      "eval_accuracy": 0.7485714285714286,
+      "eval_loss": 0.5384859442710876,
+      "eval_runtime": 26.0891,
+      "eval_samples_per_second": 6.708,
+      "eval_steps_per_second": 1.687,
       "step": 800
     },
     {
       "epoch": 0.68,
-      "learning_rate": 3.927522037218414e-05,
-      "loss": 0.593,
+      "learning_rate": 1.9686581782566114e-05,
+      "loss": 0.5231,
       "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,
+      "eval_accuracy": 0.7371428571428571,
+      "eval_loss": 0.5821840763092041,
+      "eval_runtime": 26.0996,
+      "eval_samples_per_second": 6.705,
+      "eval_steps_per_second": 1.686,
       "step": 850
     },
     {
       "epoch": 0.72,
-      "learning_rate": 3.437806072477963e-05,
-      "loss": 0.5624,
+      "learning_rate": 1.7238001958863857e-05,
+      "loss": 0.4987,
       "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,
+      "eval_accuracy": 0.7142857142857143,
+      "eval_loss": 0.5697786211967468,
+      "eval_runtime": 26.0884,
+      "eval_samples_per_second": 6.708,
+      "eval_steps_per_second": 1.687,
       "step": 900
     },
     {
       "epoch": 0.76,
-      "learning_rate": 2.9480901077375122e-05,
-      "loss": 0.5623,
+      "learning_rate": 1.4789422135161608e-05,
+      "loss": 0.5019,
       "step": 950
     },
     {
       "epoch": 0.76,
-      "eval_accuracy": 0.7371428571428571,
-      "eval_loss": 0.6078417897224426,
-      "eval_runtime": 26.2516,
-      "eval_samples_per_second": 6.666,
-      "eval_steps_per_second": 1.676,
+      "eval_accuracy": 0.7657142857142857,
+      "eval_loss": 0.5529425740242004,
+      "eval_runtime": 26.1154,
+      "eval_samples_per_second": 6.701,
+      "eval_steps_per_second": 1.685,
       "step": 950
     },
     {
       "epoch": 0.8,
-      "learning_rate": 2.458374142997062e-05,
-      "loss": 0.5835,
+      "learning_rate": 1.2340842311459353e-05,
+      "loss": 0.515,
       "step": 1000
     },
     {
       "epoch": 0.8,
       "eval_accuracy": 0.7314285714285714,
-      "eval_loss": 0.6015376448631287,
-      "eval_runtime": 28.1566,
-      "eval_samples_per_second": 6.215,
-      "eval_steps_per_second": 1.563,
+      "eval_loss": 0.571620762348175,
+      "eval_runtime": 26.0898,
+      "eval_samples_per_second": 6.708,
+      "eval_steps_per_second": 1.686,
       "step": 1000
     },
     {
       "epoch": 0.84,
-      "learning_rate": 1.9686581782566114e-05,
-      "loss": 0.5897,
+      "learning_rate": 9.8922624877571e-06,
+      "loss": 0.5154,
       "step": 1050
     },
     {
       "epoch": 0.84,
-      "eval_accuracy": 0.72,
-      "eval_loss": 0.5890141129493713,
-      "eval_runtime": 28.2155,
-      "eval_samples_per_second": 6.202,
-      "eval_steps_per_second": 1.559,
+      "eval_accuracy": 0.7257142857142858,
+      "eval_loss": 0.5566055774688721,
+      "eval_runtime": 26.1014,
+      "eval_samples_per_second": 6.705,
+      "eval_steps_per_second": 1.686,
       "step": 1050
     }
   ],
@@ -334,7 +334,7 @@
   "max_steps": 1246,
   "num_train_epochs": 1,
   "save_steps": 50,
-  "total_flos": 3.901540911082291e+17,
+  "total_flos": 3.896487034873037e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-1050/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b25ecbcf37659c2f593e1603dab54c3eaea43862c7f192a7e6df4c4dcc86d20d
+oid sha256:2b7c1677f66b49d74d914a320393e2fa29530a9670d5824a30beac9590468d16
 size 4600