<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
distilbert-base-uncased__sst2__train-8-9
This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.6925
- Accuracy: 0.5140
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: 4
- eval_batch_size: 4
- seed: 42
- 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 |
---|---|---|---|---|
0.7204 | 1.0 | 3 | 0.7025 | 0.5 |
0.6885 | 2.0 | 6 | 0.7145 | 0.5 |
0.6662 | 3.0 | 9 | 0.7222 | 0.5 |
0.6182 | 4.0 | 12 | 0.7427 | 0.25 |
0.5707 | 5.0 | 15 | 0.7773 | 0.25 |
0.5247 | 6.0 | 18 | 0.8137 | 0.25 |
0.5003 | 7.0 | 21 | 0.8556 | 0.25 |
0.4195 | 8.0 | 24 | 0.9089 | 0.5 |
0.387 | 9.0 | 27 | 0.9316 | 0.25 |
0.2971 | 10.0 | 30 | 0.9558 | 0.25 |
0.2581 | 11.0 | 33 | 0.9420 | 0.25 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2
- Tokenizers 0.10.3