deberta-v3-snall-goemotions
This model is a fine-tuned version of microsoft/deberta-v3-small on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.5638
- F1: 0.4241
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: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | F1 |
|---|---|---|---|---|
| 1.614 | 1.0 | 3082 | 1.5577 | 0.3663 |
| 1.4338 | 2.0 | 6164 | 1.5580 | 0.4084 |
| 1.2936 | 3.0 | 9246 | 1.5006 | 0.4179 |
| 1.1531 | 4.0 | 12328 | 1.5348 | 0.4276 |
| 1.0536 | 5.0 | 15410 | 1.5638 | 0.4241 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.17.0
- Tokenizers 0.10.3
- Downloads last month
- 18