<!-- 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. -->
hBERTv2_new_pretrain_48_KD_qqp
This model is a fine-tuned version of gokuls/bert_12_layer_model_v2_complete_training_new_48_KD on the GLUE QQP dataset. It achieves the following results on the evaluation set:
- Loss: 0.4160
- Accuracy: 0.8245
- F1: 0.7649
- Combined Score: 0.7947
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 | F1 | Combined Score |
---|---|---|---|---|---|---|
0.5305 | 1.0 | 2843 | 0.4755 | 0.7733 | 0.6749 | 0.7241 |
0.4434 | 2.0 | 5686 | 0.4351 | 0.7950 | 0.7260 | 0.7605 |
0.3858 | 3.0 | 8529 | 0.4173 | 0.8063 | 0.7210 | 0.7637 |
0.3409 | 4.0 | 11372 | 0.4201 | 0.7998 | 0.7579 | 0.7788 |
0.303 | 5.0 | 14215 | 0.4274 | 0.8169 | 0.7577 | 0.7873 |
0.2727 | 6.0 | 17058 | 0.4403 | 0.8186 | 0.7638 | 0.7912 |
0.2449 | 7.0 | 19901 | 0.4160 | 0.8245 | 0.7649 | 0.7947 |
0.2241 | 8.0 | 22744 | 0.4429 | 0.8271 | 0.7518 | 0.7895 |
0.2054 | 9.0 | 25587 | 0.4941 | 0.8287 | 0.7659 | 0.7973 |
0.1905 | 10.0 | 28430 | 0.4992 | 0.8264 | 0.7744 | 0.8004 |
0.1768 | 11.0 | 31273 | 0.4901 | 0.8303 | 0.7686 | 0.7995 |
0.1655 | 12.0 | 34116 | 0.5513 | 0.8276 | 0.7446 | 0.7861 |
Framework versions
- Transformers 4.30.2
- Pytorch 1.14.0a0+410ce96
- Datasets 2.12.0
- Tokenizers 0.13.3