Edit model card

fine-tuned-KoreanIndoNLI-KorNLI-with-bert-kor-base

This model is a fine-tuned version of kykim/bert-kor-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5175
  • Accuracy: 0.8032
  • F1: 0.8035

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 1e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Accuracy F1
0.513 0.5 3654 0.5627 0.7662 0.7689
0.4711 1.0 7308 0.5259 0.7822 0.7832
0.4209 1.5 10962 0.5243 0.7911 0.7907
0.4228 2.0 14616 0.4848 0.8032 0.8044
0.3934 2.5 18270 0.4994 0.8038 0.8055
0.3852 3.0 21924 0.4860 0.8089 0.8090
0.3368 3.5 25578 0.5109 0.7987 0.8001
0.3517 4.0 29232 0.4922 0.8083 0.8088
0.3072 4.5 32886 0.5175 0.8032 0.8035

Framework versions

  • Transformers 4.31.0
  • Pytorch 1.13.1
  • Datasets 2.14.4
  • Tokenizers 0.13.3
Downloads last month
23
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for muhammadravi251001/fine-tuned-KoreanIndoNLI-KorNLI-with-bert-kor-base

Finetuned
(3)
this model