metadata
library_name: transformers
language:
- np
license: mit
base_model: Sakonii/deberta-base-nepali
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
- precision
- recall
model-index:
- name: Nepali-BERT-devangari-sentiment
results: []
Nepali-BERT-devangari-sentiment
This model is a fine-tuned version of Sakonii/deberta-base-nepali on the Custom Devangari Datasets dataset. It achieves the following results on the evaluation set:
- Loss: 0.6662
- Accuracy: 0.8710
- F1: 0.5130
- Precision: 0.4571
- Recall: 0.5844
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
- lr_scheduler_warmup_steps: 500
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.6046 | 1.0 | 1189 | 0.5267 | 0.8167 | 0.4543 | 0.3475 | 0.6561 |
| 0.4952 | 2.0 | 2378 | 0.5396 | 0.8518 | 0.5025 | 0.4122 | 0.6435 |
| 0.412 | 3.0 | 3567 | 0.5733 | 0.8656 | 0.5098 | 0.4425 | 0.6013 |
| 0.3406 | 4.0 | 4756 | 0.6662 | 0.8710 | 0.5130 | 0.4571 | 0.5844 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.1+cu121
- Datasets 3.0.2
- Tokenizers 0.19.1