Training in progress, step 300, checkpoint

checkpoint-300/config.json

@@ -1,36 +1,35 @@
 {
-  "_name_or_path": "EleutherAI/polyglot-ko-1.3b",
+  "_name_or_path": "klue/roberta-large",
   "architectures": [
-    "GPTNeoXForSequenceClassification"
+    "RobertaForSequenceClassification"
   ],
-  "attention_dropout": 0.0,
+  "attention_probs_dropout_prob": 0.1,
   "bos_token_id": 0,
-  "classifier_dropout": 0.1,
+  "classifier_dropout": null,
   "eos_token_id": 2,
+  "gradient_checkpointing": false,
   "hidden_act": "gelu",
-  "hidden_dropout": 0.0,
-  "hidden_size": 2048,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
   "id2label": {
     "0": "LABEL_0"
   },
   "initializer_range": 0.02,
-  "intermediate_size": 8192,
+  "intermediate_size": 4096,
   "label2id": {
     "LABEL_0": 0
   },
   "layer_norm_eps": 1e-05,
-  "max_position_embeddings": 2048,
-  "model_type": "gpt_neox",
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
   "num_attention_heads": 16,
   "num_hidden_layers": 24,
   "pad_token_id": 2,
-  "rope_scaling": null,
-  "rotary_emb_base": 10000,
-  "rotary_pct": 0.5,
-  "tie_word_embeddings": false,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "BertTokenizer",
   "torch_dtype": "float32",
   "transformers_version": "4.34.1",
+  "type_vocab_size": 1,
   "use_cache": true,
-  "use_parallel_residual": true,
-  "vocab_size": 30080
+  "vocab_size": 32000
 }

checkpoint-300/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba1c2998331a8d59c9e4fd0e95ca45b9e7307dbe5662c1fa52581a901dc63590
-size 10161845830
+oid sha256:b115d8db8fff140c76acfc901505a07de0f55a1a7e5ea897df796de9dbe555e9
+size 2693490618

checkpoint-300/pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8c94c5e1ded3e3eb3775e169b4b2921545c157a5fa927d7b4ba16f2ec3497954
-size 5080945162
+oid sha256:2a29432812ebb1bc36254133e6052c46003ff6f255f77850b1e6b443c21ef9e9
+size 1346765678

checkpoint-300/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1dbca245f4c6603645dbab7224b3a520ea557088259edf5fb8205675ece2a192
+oid sha256:aa36aa7f15c04af6a625682b18c55ce92517a9540df54f7516a6beb884cbc3b4
 size 14308

checkpoint-300/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:153655dd49971685f6e46e64b8fe8ae819c331baf597a84fccbaf9b90e0354b0
+oid sha256:edcc4d73ffd0cebeb2e36129b0d167522d159cf15f1c212ba2d19a320db063de
 size 1064

checkpoint-300/special_tokens_map.json

@@ -1,11 +1,9 @@
 {
-  "additional_special_tokens": [
-    "<|endoftext|>",
-    "<|sep|>",
-    "<|acc|>",
-    "<|tel|>",
-    "<|rrn|>"
-  ],
-  "eos_token": "<|endoftext|>",
-  "pad_token": "<|endoftext|>"
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
 }

checkpoint-300/tokenizer_config.json

@@ -1,7 +1,7 @@
 {
   "added_tokens_decoder": {
     "0": {
-      "content": "<|unused0|>",
+      "content": "[CLS]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -9,7 +9,7 @@
       "special": true
     },
     "1": {
-      "content": "<|unused1|>",
+      "content": "[PAD]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -17,7 +17,7 @@
       "special": true
     },
     "2": {
-      "content": "<|endoftext|>",
+      "content": "[SEP]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -25,31 +25,15 @@
       "special": true
     },
     "3": {
-      "content": "<|sep|>",
+      "content": "[UNK]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "30000": {
-      "content": "<|acc|>",
-      "lstrip": false,
-      "normalized": false,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    },
-    "30001": {
-      "content": "<|tel|>",
-      "lstrip": false,
-      "normalized": false,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    },
-    "30002": {
-      "content": "<|rrn|>",
+    "4": {
+      "content": "[MASK]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -57,16 +41,19 @@
       "special": true
     }
   },
-  "additional_special_tokens": [
-    "<|endoftext|>",
-    "<|sep|>",
-    "<|acc|>",
-    "<|tel|>",
-    "<|rrn|>"
-  ],
+  "bos_token": "[CLS]",
   "clean_up_tokenization_spaces": true,
-  "eos_token": "<|endoftext|>",
-  "model_max_length": 1000000000000000019884624838656,
-  "pad_token": "<|endoftext|>",
-  "tokenizer_class": "PreTrainedTokenizerFast"
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "unk_token": "[UNK]"
 }

checkpoint-300/trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.4877057508636456,
+  "epoch": 0.45724737082761774,
   "eval_steps": 100,
   "global_step": 300,
   "is_hyper_param_search": false,
@@ -10,214 +10,214 @@
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 8.994130006901536e-06,
-      "loss": 0.3053,
+      "learning_rate": 8.99484069051768e-06,
+      "loss": 0.672,
       "step": 10
     },
     {
       "epoch": 0.03,
-      "learning_rate": 8.976535341747906e-06,
-      "loss": 0.1363,
+      "learning_rate": 8.979374592503753e-06,
+      "loss": 0.393,
       "step": 20
     },
     {
       "epoch": 0.05,
-      "learning_rate": 8.947261907011569e-06,
-      "loss": 0.1087,
+      "learning_rate": 8.953637170129838e-06,
+      "loss": 0.1594,
       "step": 30
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 8.906386073741352e-06,
-      "loss": 0.106,
+      "epoch": 0.06,
+      "learning_rate": 8.917687439985847e-06,
+      "loss": 0.122,
       "step": 40
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.85401448231912e-06,
-      "loss": 0.0791,
+      "learning_rate": 8.87160783575339e-06,
+      "loss": 0.0956,
       "step": 50
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 8.790283764247187e-06,
-      "loss": 0.0702,
+      "epoch": 0.09,
+      "learning_rate": 8.815504019183158e-06,
+      "loss": 0.102,
       "step": 60
     },
     {
       "epoch": 0.11,
-      "learning_rate": 8.715360185692326e-06,
-      "loss": 0.0773,
+      "learning_rate": 8.749504637809742e-06,
+      "loss": 0.0888,
       "step": 70
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 8.629439213716327e-06,
-      "loss": 0.0583,
+      "epoch": 0.12,
+      "learning_rate": 8.673761029959426e-06,
+      "loss": 0.097,
       "step": 80
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 8.532745006324751e-06,
-      "loss": 0.0637,
+      "epoch": 0.14,
+      "learning_rate": 8.588446877727417e-06,
+      "loss": 0.0968,
       "step": 90
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 8.425529827664287e-06,
-      "loss": 0.0554,
+      "epoch": 0.15,
+      "learning_rate": 8.493757808720196e-06,
+      "loss": 0.0808,
       "step": 100
     },
     {
-      "epoch": 0.16,
-      "eval_accuracy": 0.5,
-      "eval_loss": 7.870811641330633e-11,
-      "eval_runtime": 51.3409,
-      "eval_samples_per_second": 85.195,
-      "eval_steps_per_second": 10.654,
+      "epoch": 0.15,
+      "eval_accuracy": 0.0,
+      "eval_loss": 3.0685205274494365e-05,
+      "eval_runtime": 30.877,
+      "eval_samples_per_second": 141.659,
+      "eval_steps_per_second": 23.61,
       "step": 100
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 8.308073389894399e-06,
-      "loss": 0.0666,
+      "epoch": 0.17,
+      "learning_rate": 8.389910947476252e-06,
+      "loss": 0.0797,
       "step": 110
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 8.18068212345023e-06,
-      "loss": 0.0598,
+      "epoch": 0.18,
+      "learning_rate": 8.277144417593776e-06,
+      "loss": 0.0933,
       "step": 120
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 8.043688377600596e-06,
-      "loss": 0.0423,
+      "epoch": 0.2,
+      "learning_rate": 8.155716795706948e-06,
+      "loss": 0.084,
       "step": 130
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 7.897449553386683e-06,
-      "loss": 0.0311,
+      "epoch": 0.21,
+      "learning_rate": 8.02590651856289e-06,
+      "loss": 0.0605,
       "step": 140
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 7.742347171203542e-06,
-      "loss": 0.0527,
+      "epoch": 0.23,
+      "learning_rate": 7.88801124455883e-06,
+      "loss": 0.0746,
       "step": 150
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 7.578785875456937e-06,
-      "loss": 0.0847,
+      "epoch": 0.24,
+      "learning_rate": 7.742347171203542e-06,
+      "loss": 0.1016,
       "step": 160
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 7.407192378892295e-06,
-      "loss": 0.0595,
+      "epoch": 0.26,
+      "learning_rate": 7.589248310068091e-06,
+      "loss": 0.0919,
       "step": 170
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 7.2280143493498716e-06,
-      "loss": 0.0327,
+      "epoch": 0.27,
+      "learning_rate": 7.429065720888471e-06,
+      "loss": 0.1076,
       "step": 180
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 7.041719241850471e-06,
-      "loss": 0.0877,
+      "epoch": 0.29,
+      "learning_rate": 7.262166706576352e-06,
+      "loss": 0.0586,
       "step": 190
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 6.848793079058672e-06,
-      "loss": 0.0356,
+      "epoch": 0.3,
+      "learning_rate": 7.088933970983793e-06,
+      "loss": 0.0682,
       "step": 200
     },
     {
-      "epoch": 0.33,
-      "eval_accuracy": 0.5,
-      "eval_loss": 9.303278147854144e-08,
-      "eval_runtime": 53.2016,
-      "eval_samples_per_second": 82.216,
-      "eval_steps_per_second": 10.282,
+      "epoch": 0.3,
+      "eval_accuracy": 0.0,
+      "eval_loss": 9.85040060186293e-06,
+      "eval_runtime": 31.5055,
+      "eval_samples_per_second": 138.833,
+      "eval_steps_per_second": 23.139,
       "step": 200
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 6.649739183305183e-06,
-      "loss": 0.0332,
+      "epoch": 0.32,
+      "learning_rate": 6.909764741353188e-06,
+      "loss": 0.069,
       "step": 210
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 6.445076863476361e-06,
-      "loss": 0.0537,
+      "epoch": 0.34,
+      "learning_rate": 6.7250698574647064e-06,
+      "loss": 0.0483,
       "step": 220
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 6.235340060196612e-06,
-      "loss": 0.0548,
+      "epoch": 0.35,
+      "learning_rate": 6.5352728295698045e-06,
+      "loss": 0.0639,
       "step": 230
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 6.021075952838263e-06,
-      "loss": 0.0485,
+      "epoch": 0.37,
+      "learning_rate": 6.340808867271031e-06,
+      "loss": 0.0865,
       "step": 240
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 5.802843531993069e-06,
-      "loss": 0.041,
+      "epoch": 0.38,
+      "learning_rate": 6.142123881574899e-06,
+      "loss": 0.0861,
       "step": 250
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 5.581212141129573e-06,
-      "loss": 0.0266,
+      "epoch": 0.4,
+      "learning_rate": 5.939673462406165e-06,
+      "loss": 0.0805,
       "step": 260
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 5.3567599912410075e-06,
-      "loss": 0.0421,
+      "epoch": 0.41,
+      "learning_rate": 5.7339218339280815e-06,
+      "loss": 0.0751,
       "step": 270
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 5.130072652358832e-06,
-      "loss": 0.0522,
+      "epoch": 0.43,
+      "learning_rate": 5.52534079006412e-06,
+      "loss": 0.0687,
       "step": 280
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 4.901741525867374e-06,
-      "loss": 0.0302,
+      "epoch": 0.44,
+      "learning_rate": 5.314408612662028e-06,
+      "loss": 0.092,
       "step": 290
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.67236230160516e-06,
-      "loss": 0.0741,
+      "epoch": 0.46,
+      "learning_rate": 5.101608974780918e-06,
+      "loss": 0.0718,
       "step": 300
     },
     {
-      "epoch": 0.49,
+      "epoch": 0.46,
       "eval_accuracy": 0.0,
-      "eval_loss": 2.2148276457301108e-07,
-      "eval_runtime": 51.7845,
-      "eval_samples_per_second": 84.465,
-      "eval_steps_per_second": 10.563,
+      "eval_loss": 7.04687363395351e-06,
+      "eval_runtime": 30.7873,
+      "eval_samples_per_second": 142.071,
+      "eval_steps_per_second": 23.679,
       "step": 300
     }
   ],
   "logging_steps": 10,
-  "max_steps": 615,
+  "max_steps": 656,
   "num_train_epochs": 1,
   "save_steps": 100,
   "total_flos": 0.0,

checkpoint-300/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4f83b04586007121ba3ee6bfcd4a06860c6098884f4f89ada67554e532440c28
+oid sha256:e1b0be5f447301b577e44061c504ec9398fb05913a538169784e7cfc7134c7d3
 size 4600