small-vanilla-target-glue-qqp
This model is a fine-tuned version of google/bert_uncased_L-4_H-512_A-8 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3259
- Accuracy: 0.8536
- F1: 0.8135
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: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- training_steps: 5000
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 0.4725 | 0.04 | 500 | 0.4206 | 0.7941 | 0.7492 |
| 0.4177 | 0.09 | 1000 | 0.3860 | 0.8157 | 0.7712 |
| 0.4009 | 0.13 | 1500 | 0.3750 | 0.8221 | 0.7756 |
| 0.3815 | 0.18 | 2000 | 0.3781 | 0.8206 | 0.7858 |
| 0.3768 | 0.22 | 2500 | 0.3495 | 0.8395 | 0.7975 |
| 0.3592 | 0.26 | 3000 | 0.3496 | 0.8379 | 0.7993 |
| 0.359 | 0.31 | 3500 | 0.3291 | 0.8514 | 0.8020 |
| 0.3504 | 0.35 | 4000 | 0.3342 | 0.8469 | 0.8057 |
| 0.3582 | 0.4 | 4500 | 0.3345 | 0.8482 | 0.8095 |
| 0.3434 | 0.44 | 5000 | 0.3259 | 0.8536 | 0.8135 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu116
- Datasets 2.8.1.dev0
- Tokenizers 0.13.2