bert_uncased_L-10_H-256_A-4_massive
This model is a fine-tuned version of google/bert_uncased_L-10_H-256_A-4 on the massive dataset. It achieves the following results on the evaluation set:
- Loss: 0.6673
- Accuracy: 0.8539
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: 15
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 3.3641 | 1.0 | 180 | 2.6029 | 0.5519 |
| 2.2711 | 2.0 | 360 | 1.8561 | 0.7029 |
| 1.68 | 3.0 | 540 | 1.4348 | 0.7545 |
| 1.3057 | 4.0 | 720 | 1.1809 | 0.7905 |
| 1.0483 | 5.0 | 900 | 1.0132 | 0.8106 |
| 0.8633 | 6.0 | 1080 | 0.9081 | 0.8239 |
| 0.7338 | 7.0 | 1260 | 0.8341 | 0.8298 |
| 0.6226 | 8.0 | 1440 | 0.7731 | 0.8401 |
| 0.5447 | 9.0 | 1620 | 0.7396 | 0.8406 |
| 0.4824 | 10.0 | 1800 | 0.7151 | 0.8490 |
| 0.4354 | 11.0 | 1980 | 0.7003 | 0.8495 |
| 0.3977 | 12.0 | 2160 | 0.6864 | 0.8500 |
| 0.3723 | 13.0 | 2340 | 0.6753 | 0.8529 |
| 0.3543 | 14.0 | 2520 | 0.6740 | 0.8500 |
| 0.3409 | 15.0 | 2700 | 0.6673 | 0.8539 |
Framework versions
- Transformers 4.34.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.14.5
- Tokenizers 0.14.1