Hyeonseo's picture
update model card README.md
d1ae4c3
|
raw
history blame
2.56 kB
metadata
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: ko-finance_news_classifier
    results: []

ko-finance_news_classifier

This model is a fine-tuned version of cardiffnlp/twitter-xlm-roberta-base-sentiment on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.4474
  • Accuracy: 0.8423

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: 2e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 243 1.0782 0.8010
No log 2.0 486 1.0328 0.8381
0.0766 3.0 729 1.2348 0.8330
0.0766 4.0 972 1.3915 0.8052
0.046 5.0 1215 1.2995 0.8474
0.046 6.0 1458 1.2926 0.8361
0.0512 7.0 1701 1.2889 0.8330
0.0512 8.0 1944 1.3107 0.8392
0.0415 9.0 2187 1.4514 0.8309
0.0415 10.0 2430 1.2869 0.8381
0.0279 11.0 2673 1.2874 0.8526
0.0279 12.0 2916 1.4731 0.8423
0.0126 13.0 3159 1.3956 0.8443
0.0126 14.0 3402 1.4211 0.8454
0.0101 15.0 3645 1.3686 0.8474
0.0101 16.0 3888 1.4412 0.8423
0.0114 17.0 4131 1.4376 0.8423
0.0114 18.0 4374 1.4566 0.8423
0.0055 19.0 4617 1.4439 0.8443
0.0055 20.0 4860 1.4474 0.8423

Framework versions

  • Transformers 4.28.0
  • Pytorch 2.0.1+cu118
  • Datasets 2.12.0
  • Tokenizers 0.13.3