Training in progress, step 400, checkpoint

checkpoint-400/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-400/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9d4ecf8ea5ff41e71b2db5cf9b06c3edb02ba6fb9cbec652801d040632f7547d
-size 9859800
+oid sha256:be0f7ffbcecdb761c4fc217f7fd68f6666f2ce319ce20b8dcd5dc80c462fdbc6
+size 19690328

checkpoint-400/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:657d1f865cbfe32985cc45deead2387554ec36c0fb48969506bfe4d2a9b9fe8f
-size 42788
+oid sha256:1dcc5d183f0267932b5d7abf1f0cb90952ab502a99189fdf95d0442aec91c64d
+size 38087162

checkpoint-400/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d817a3cb9b4a41f5690417c1696c49d5b09b444b7851d7934ef29b003995b1f6
+oid sha256:a86072e57b8bf8f2c9b2bfac963bc3871c76f45cfa678f1c3ef9b3a9f74c7d2b
 size 14244

checkpoint-400/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:38c429b68246a9d96d6dbdf90e931cb2ad40f276c7acb51e3a7c3536b820795a
+oid sha256:2ffa4da8986ef90afe2d6da0c026fe3eaeef42866505d146a0f81858bdd80f65
 size 1064

checkpoint-400/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-400/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-400/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.522564172744751,
-  "best_model_checkpoint": "output/checkpoint-50",
-  "epoch": 2.7923211169284468,
+  "best_metric": 0.7192752361297607,
+  "best_model_checkpoint": "output/checkpoint-400",
+  "epoch": 2.04211869814933,
   "eval_steps": 50,
   "global_step": 400,
   "is_hyper_param_search": false,
@@ -10,136 +10,136 @@
   "log_history": [
     {
       "epoch": 0.01,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 0.4196,
+      "learning_rate": 2.777777777777778e-07,
+      "loss": 9.5569,
       "step": 1
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.6e-06,
-      "loss": 0.4279,
+      "epoch": 0.26,
+      "learning_rate": 1.25e-05,
+      "loss": 6.958,
       "step": 50
     },
     {
-      "epoch": 0.35,
-      "eval_accuracy": 0.7657142857142857,
-      "eval_loss": 0.522564172744751,
-      "eval_runtime": 29.0326,
-      "eval_samples_per_second": 6.028,
-      "eval_steps_per_second": 1.516,
+      "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": 0.7,
-      "learning_rate": 8.86829268292683e-06,
-      "loss": 0.4089,
+      "epoch": 0.51,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 1.6141,
       "step": 100
     },
     {
-      "epoch": 0.7,
-      "eval_accuracy": 0.7428571428571429,
-      "eval_loss": 0.5401451587677002,
-      "eval_runtime": 29.0003,
-      "eval_samples_per_second": 6.034,
-      "eval_steps_per_second": 1.517,
+      "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": 1.05,
-      "learning_rate": 7.792682926829268e-06,
-      "loss": 0.4018,
+      "epoch": 0.77,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.8533,
       "step": 150
     },
     {
-      "epoch": 1.05,
-      "eval_accuracy": 0.7428571428571429,
-      "eval_loss": 0.5425500273704529,
-      "eval_runtime": 28.9813,
-      "eval_samples_per_second": 6.038,
-      "eval_steps_per_second": 1.518,
+      "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": 1.4,
-      "learning_rate": 6.695121951219512e-06,
-      "loss": 0.41,
+      "epoch": 1.02,
+      "learning_rate": 4.908536585365854e-05,
+      "loss": 0.4474,
       "step": 200
     },
     {
-      "epoch": 1.4,
-      "eval_accuracy": 0.7542857142857143,
-      "eval_loss": 0.5254732966423035,
-      "eval_runtime": 28.8187,
-      "eval_samples_per_second": 6.072,
-      "eval_steps_per_second": 1.527,
+      "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": 1.75,
-      "learning_rate": 5.597560975609756e-06,
-      "loss": 0.3927,
+      "epoch": 1.28,
+      "learning_rate": 4.603658536585366e-05,
+      "loss": 0.2996,
       "step": 250
     },
     {
-      "epoch": 1.75,
-      "eval_accuracy": 0.7485714285714286,
-      "eval_loss": 0.5406104326248169,
-      "eval_runtime": 27.5585,
-      "eval_samples_per_second": 6.35,
-      "eval_steps_per_second": 1.597,
+      "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": 2.09,
-      "learning_rate": 4.5e-06,
-      "loss": 0.428,
+      "epoch": 1.53,
+      "learning_rate": 4.298780487804878e-05,
+      "loss": 0.2403,
       "step": 300
     },
     {
-      "epoch": 2.09,
-      "eval_accuracy": 0.76,
-      "eval_loss": 0.5347301363945007,
-      "eval_runtime": 28.9003,
-      "eval_samples_per_second": 6.055,
-      "eval_steps_per_second": 1.522,
+      "epoch": 1.53,
+      "eval_accuracy": 0.72,
+      "eval_loss": 1.076162338256836,
+      "eval_runtime": 26.2142,
+      "eval_samples_per_second": 6.676,
+      "eval_steps_per_second": 1.678,
       "step": 300
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 3.402439024390244e-06,
-      "loss": 0.403,
+      "epoch": 1.79,
+      "learning_rate": 3.9939024390243905e-05,
+      "loss": 0.1982,
       "step": 350
     },
     {
-      "epoch": 2.44,
-      "eval_accuracy": 0.7428571428571429,
-      "eval_loss": 0.5431011319160461,
-      "eval_runtime": 26.5045,
-      "eval_samples_per_second": 6.603,
-      "eval_steps_per_second": 1.66,
+      "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": 2.79,
-      "learning_rate": 2.304878048780488e-06,
-      "loss": 0.4024,
+      "epoch": 2.04,
+      "learning_rate": 3.6890243902439025e-05,
+      "loss": 0.1926,
       "step": 400
     },
     {
-      "epoch": 2.79,
-      "eval_accuracy": 0.7485714285714286,
-      "eval_loss": 0.5433735847473145,
-      "eval_runtime": 26.4908,
-      "eval_samples_per_second": 6.606,
-      "eval_steps_per_second": 1.661,
+      "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
     }
   ],
   "logging_steps": 50,
-  "max_steps": 500,
-  "num_train_epochs": 4,
+  "max_steps": 1000,
+  "num_train_epochs": 6,
   "save_steps": 50,
-  "total_flos": 1.502465282525184e+17,
+  "total_flos": 1.48981821848064e+17,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-400/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b1b0a94354c0f3dfee9c93e225dbd48ad6f856f98977ef38ff3415910328e64a
+oid sha256:5a0ef36d7c31265f68a3edc8ae53ce9272dcc90a79f261e29a2a46e040c77a33
 size 4600