soonbob/mnli-finetuned-bert-base-cased

MNLI 데이터셋을 학습시킨 BERT 파인튜닝 연습용으로 만든 것입니다.

This is a BERT-based model fine-tuned on the Multi-Genre Natural Language Inference (MultiNLI) dataset for the task of natural language inference (NLI), using Hugging Face's Trainer.

It classifies a pair of sentences into one of the following classes:

• entailment
• neutral
• contradiction

🧠 Intended Use

This model can be used for:

• Evaluating whether one sentence logically follows from another
• Sentence-pair classification tasks
• Transfer learning for other NLI-style problems

It achieves the following results on the evaluation set:

• Loss: 0.8276
• Accuracy: 0.6368
• F1: 0.6358

⚙️ Training Details

• Base model: bert-base-cased
• Dataset: nyu-mll/glue, subset: mnli
• Epochs: 3
• Learning rate: 1e-3
• Optimizer: AdamW
• Scheduler: Linear

🏋️ Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.001
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

🏋️ Training Logs

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.8662	1.0	2455	0.8682	0.6033	0.5946
0.7964	2.0	4910	0.8449	0.6242	0.6242
0.7323	3.0	7365	0.8673	0.6237	0.6231

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1