feat: Fine-tune RoBERTa-large with KorQuAD v1

README.md

@@ -9,7 +9,7 @@ tags:
 ---
 
 ### Config
-- epoch = 2
+- epoch = 1
 - learning_rate = 1e-5
 - batch_size = 16
 
@@ -17,5 +17,4 @@ tags:
 - V100(RAM 32GB)
 
 ### Validation Result
-- epoch = 1: `EM` 88.1191, `F1` 92.8047
-- epoch = 2: `EM` , `F1` 
+- epoch = 1: `EM` 88.1191, `F1` 92.8047

all_results.json

@@ -0,0 +1,12 @@
+{
+    "epoch": 1.0,
+    "eval_samples": 6865,
+    "exact_match": 88.11915483200555,
+    "f1": 92.80469400706696,
+    "total_flos": 4.801972373497037e+16,
+    "train_loss": 0.6182623529500156,
+    "train_runtime": 1876.03,
+    "train_samples": 69399,
+    "train_samples_per_second": 36.992,
+    "train_steps_per_second": 2.312
+}

config.json

@@ -0,0 +1,29 @@
+{
+  "_name_or_path": "klue/roberta-large",
+  "architectures": [
+    "RobertaForQuestionAnswering"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "BertTokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.45.2",
+  "type_vocab_size": 1,
+  "use_cache": true,
+  "vocab_size": 32000
+}

eval_results.json

@@ -0,0 +1,6 @@
+{
+    "epoch": 1.0,
+    "eval_samples": 6865,
+    "exact_match": 88.11915483200555,
+    "f1": 92.80469400706696
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0aa4d93967c0fc1ccae387f0a6872064382ede5aa14747558fb53e198dec306c
+size 1342483296

special_tokens_map.json

@@ -0,0 +1,51 @@
+{
+  "bos_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "cls_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "[MASK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,59 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "[CLS]",
+  "clean_up_tokenization_spaces": false,
+  "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]"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 1.0,
+    "total_flos": 4.801972373497037e+16,
+    "train_loss": 0.6182623529500156,
+    "train_runtime": 1876.03,
+    "train_samples": 69399,
+    "train_samples_per_second": 36.992,
+    "train_steps_per_second": 2.312
+}

train_results.txt

@@ -0,0 +1,7 @@
+epoch = 1.0
+total_flos = 4.801972373497037e+16
+train_loss = 0.6182623529500156
+train_runtime = 1876.03
+train_samples = 69399
+train_samples_per_second = 36.992
+train_steps_per_second = 2.312

trainer_state.json

@@ -0,0 +1,98 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 50,
+  "global_step": 4338,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.11526048870447211,
+      "grad_norm": 20.16172218322754,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.3764,
+      "step": 500
+    },
+    {
+      "epoch": 0.23052097740894423,
+      "grad_norm": 16.672468185424805,
+      "learning_rate": 8.705054715997916e-06,
+      "loss": 0.5451,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3457814661134163,
+      "grad_norm": 11.151906967163086,
+      "learning_rate": 7.402292860865034e-06,
+      "loss": 0.4302,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46104195481788846,
+      "grad_norm": 7.392632961273193,
+      "learning_rate": 6.099531005732152e-06,
+      "loss": 0.3772,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5763024435223605,
+      "grad_norm": 6.710408687591553,
+      "learning_rate": 4.796769150599271e-06,
+      "loss": 0.3742,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6915629322268326,
+      "grad_norm": 8.172633171081543,
+      "learning_rate": 3.494007295466389e-06,
+      "loss": 0.3518,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8068234209313048,
+      "grad_norm": 5.664760112762451,
+      "learning_rate": 2.1912454403335073e-06,
+      "loss": 0.3619,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9220839096357769,
+      "grad_norm": 16.45347785949707,
+      "learning_rate": 8.884835852006255e-07,
+      "loss": 0.3293,
+      "step": 4000
+    },
+    {
+      "epoch": 1.0,
+      "step": 4338,
+      "total_flos": 4.801972373497037e+16,
+      "train_loss": 0.6182623529500156,
+      "train_runtime": 1876.03,
+      "train_samples_per_second": 36.992,
+      "train_steps_per_second": 2.312
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 4338,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.801972373497037e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79abd64a81d1628797e3b05d617cbcfd462e49d3557e35556bd00b2bfba490ec
+size 4731