small-vanilla-target-glue-mnli-linear-probe
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: 1.0612
- Accuracy: 0.4363
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 |
|---|---|---|---|---|
| 1.1093 | 0.04 | 500 | 1.0875 | 0.3914 |
| 1.089 | 0.08 | 1000 | 1.0814 | 0.3988 |
| 1.0811 | 0.12 | 1500 | 1.0760 | 0.4113 |
| 1.0753 | 0.16 | 2000 | 1.0728 | 0.4200 |
| 1.0758 | 0.2 | 2500 | 1.0702 | 0.4252 |
| 1.0727 | 0.24 | 3000 | 1.0684 | 0.4269 |
| 1.0707 | 0.29 | 3500 | 1.0665 | 0.4295 |
| 1.0702 | 0.33 | 4000 | 1.0648 | 0.4317 |
| 1.0654 | 0.37 | 4500 | 1.0627 | 0.4352 |
| 1.0637 | 0.41 | 5000 | 1.0612 | 0.4363 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu116
- Datasets 2.8.1.dev0
- Tokenizers 0.13.2