indic-transformers-te-distilbert
This model was trained from scratch on the wikiann dataset. It achieves the following results on the evaluation set:
- Loss: 0.2940
- Precision: 0.5657
- Recall: 0.6486
- F1: 0.6043
- Accuracy: 0.9049
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
No log | 1.0 | 125 | 0.3629 | 0.4855 | 0.5287 | 0.5062 | 0.8826 |
No log | 2.0 | 250 | 0.3032 | 0.5446 | 0.6303 | 0.5843 | 0.9002 |
No log | 3.0 | 375 | 0.2940 | 0.5657 | 0.6486 | 0.6043 | 0.9049 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.17.0
- Tokenizers 0.10.3
- Downloads last month
- 121
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
Dataset used to train durgaamma2005/indic-transformers-te-distilbert
Evaluation results
- Precision on wikiannself-reported0.566
- Recall on wikiannself-reported0.649
- F1 on wikiannself-reported0.604
- Accuracy on wikiannself-reported0.905