hBERTv1_new_pretrain_w_init__qnli
This model is a fine-tuned version of gokuls/bert_12_layer_model_v1_complete_training_new_wt_init on the GLUE QNLI dataset. It achieves the following results on the evaluation set:
- Loss: 0.6672
- Accuracy: 0.5986
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: 4e-05
- train_batch_size: 128
- eval_batch_size: 128
- 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
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.6909 | 1.0 | 819 | 0.6783 | 0.5653 |
| 0.684 | 2.0 | 1638 | 0.6904 | 0.5100 |
| 0.6765 | 3.0 | 2457 | 0.6709 | 0.5881 |
| 0.6696 | 4.0 | 3276 | 0.6774 | 0.5695 |
| 0.6676 | 5.0 | 4095 | 0.6704 | 0.5903 |
| 0.6626 | 6.0 | 4914 | 0.6672 | 0.5986 |
| 0.6661 | 7.0 | 5733 | 0.6703 | 0.5907 |
| 0.6642 | 8.0 | 6552 | 0.6693 | 0.5960 |
| 0.6698 | 9.0 | 7371 | 0.6733 | 0.5799 |
| 0.6724 | 10.0 | 8190 | 0.6815 | 0.5636 |
| 0.68 | 11.0 | 9009 | 0.6908 | 0.5427 |
Framework versions
- Transformers 4.29.2
- Pytorch 1.14.0a0+410ce96
- Datasets 2.12.0
- Tokenizers 0.13.3