bert_uncased_L-6_H-256_A-4_emotion
This model is a fine-tuned version of google/bert_uncased_L-6_H-256_A-4 on the emotion dataset. It achieves the following results on the evaluation set:
- Loss: 0.1706
- Accuracy: 0.938
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: 5e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 33
- distributed_type: multi-GPU
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.0209 | 1.0 | 250 | 0.4902 | 0.872 |
| 0.3573 | 2.0 | 500 | 0.2427 | 0.9235 |
| 0.2124 | 3.0 | 750 | 0.1885 | 0.9295 |
| 0.1605 | 4.0 | 1000 | 0.1815 | 0.9335 |
| 0.137 | 5.0 | 1250 | 0.1623 | 0.9355 |
| 0.1122 | 6.0 | 1500 | 0.1695 | 0.934 |
| 0.0968 | 7.0 | 1750 | 0.1671 | 0.935 |
| 0.0902 | 8.0 | 2000 | 0.1702 | 0.933 |
| 0.08 | 9.0 | 2250 | 0.1684 | 0.937 |
| 0.0724 | 10.0 | 2500 | 0.1706 | 0.938 |
Framework versions
- Transformers 4.34.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.14.5
- Tokenizers 0.14.1