distilbert_sa_GLUE_Experiment_mnli_256
This model is a fine-tuned version of distilbert-base-uncased on the GLUE MNLI dataset. It achieves the following results on the evaluation set:
- Loss: 0.8831
- Accuracy: 0.5886
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: 256
- eval_batch_size: 256
- seed: 10
- distributed_type: multi-GPU
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.0194 | 1.0 | 1534 | 0.9641 | 0.5250 |
| 0.9428 | 2.0 | 3068 | 0.9256 | 0.5586 |
| 0.9042 | 3.0 | 4602 | 0.9137 | 0.5684 |
| 0.8725 | 4.0 | 6136 | 0.8912 | 0.5849 |
| 0.8471 | 5.0 | 7670 | 0.8784 | 0.5930 |
| 0.8242 | 6.0 | 9204 | 0.8841 | 0.5932 |
| 0.8037 | 7.0 | 10738 | 0.8826 | 0.6006 |
| 0.784 | 8.0 | 12272 | 0.9013 | 0.5946 |
| 0.7647 | 9.0 | 13806 | 0.8934 | 0.6054 |
| 0.7468 | 10.0 | 15340 | 0.8993 | 0.6042 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.8.0
- Tokenizers 0.13.2