lora-midm-nsmc

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

lora-midm-nsmc

This model is a fine-tuned version of KT-AI/midm-bitext-S-7B-inst-v1 on an nsmc dataset.

Model description

KT-midm model을 nsmc데이터를 활용하여 미세튜닝한 모델
영화 리뷰 데이터를 기반으로 사용자가 작성한 리뷰의 긍정 또는 부정을 파악한다.

Intended uses & limitations

Intended uses

사용자가 작성한 리뷰의 긍정 또는 부정 감정 분석을 제공함

Limitaions

영화 리뷰에 특화되어 있으며, 다른 유형에는 제한이 있을 수 있음
Colab T4 GPU에서 테스트 되었음

Training and evaluation data

Training data: nsmc 'train' data 중 상위 2000개의 샘플
Evaluation data: nsmc 'test' data 중 상위 1000개의 샘플

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 2
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• training_steps: 300
• mixed_precision_training: Native AMP

Training results

TrainOutput(global_step=300, training_loss=1.1105608495076498,
metrics={'train_runtime': 929.3252, 'train_samples_per_second': 0.646,
'train_steps_per_second': 0.323, 'total_flos': 9315508499251200.0,
'train_loss': 1.1105608495076498, 'epoch': 0.3})

정확도

Midm: 정확도 0.89

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0