Ransaka's picture
Rename model
074501e
|
raw
history blame
2.52 kB
metadata
license: mit
base_model: Ransaka/sinhala-bert-medium-v2
tags:
  - generated_from_trainer
metrics:
  - f1
model-index:
  - name: SentimentClassifier.si
    results: []
language:
  - si
pipeline_tag: text-classification

SentimentClassifier.si

This model is a fine-tuned version of Ransaka/sinhala-bert-medium-v2 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2358
  • F1: 0.8877

Intended uses & limitations

More information needed

Training and evaluation data

Labels

  NEGATIVE: 1
  POSITIVE: 0

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.0002
  • 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
  • training_steps: 1000

Training results

Training Loss Epoch Step Validation Loss F1
0.4053 0.08 100 0.2802 0.8677
0.3768 0.16 200 0.3123 0.8616
0.3334 0.24 300 0.2810 0.8732
0.2906 0.32 400 0.2554 0.8779
0.3027 0.4 500 0.2595 0.8836
0.2612 0.48 600 0.2797 0.8592
0.2568 0.56 700 0.2474 0.8785
0.2325 0.64 800 0.2546 0.8816
0.2272 0.72 900 0.2424 0.8878
0.2331 0.8 1000 0.2358 0.8877

Model performance on validation dataset

              precision    recall  f1-score   support

           0       0.95      0.92      0.93      6943
           1       0.82      0.88      0.84      2913

    accuracy                           0.90      9856
   macro avg       0.88      0.90      0.89      9856
weighted avg       0.91      0.90      0.91      9856

Confusion Matrix on Validation Data

Framework versions

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