<!-- 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_sa_GLUE_Experiment_logit_kd_qqp_256
This model is a fine-tuned version of distilbert-base-uncased on the GLUE QQP dataset. It achieves the following results on the evaluation set:
- Loss: 0.6955
- Accuracy: 0.6386
- F1: 0.0394
- Combined Score: 0.3390
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: 5e-05
- train_batch_size: 256
- eval_batch_size: 256
- 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
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Combined Score |
---|---|---|---|---|---|---|
0.8148 | 1.0 | 1422 | 0.7743 | 0.6318 | 0.0 | 0.3159 |
0.7531 | 2.0 | 2844 | 0.7520 | 0.6318 | 0.0 | 0.3159 |
0.7248 | 3.0 | 4266 | 0.7365 | 0.6318 | 0.0 | 0.3159 |
0.6964 | 4.0 | 5688 | 0.7183 | 0.6444 | 0.0751 | 0.3598 |
0.676 | 5.0 | 7110 | 0.7078 | 0.6346 | 0.0165 | 0.3256 |
0.6579 | 6.0 | 8532 | 0.7030 | 0.6359 | 0.0241 | 0.3300 |
0.6427 | 7.0 | 9954 | 0.7007 | 0.6332 | 0.0078 | 0.3205 |
0.6291 | 8.0 | 11376 | 0.7044 | 0.6389 | 0.0415 | 0.3402 |
0.6173 | 9.0 | 12798 | 0.7002 | 0.6407 | 0.0522 | 0.3465 |
0.6061 | 10.0 | 14220 | 0.6971 | 0.6406 | 0.0526 | 0.3466 |
0.5967 | 11.0 | 15642 | 0.6955 | 0.6386 | 0.0394 | 0.3390 |
0.5871 | 12.0 | 17064 | 0.7006 | 0.6470 | 0.0874 | 0.3672 |
0.5792 | 13.0 | 18486 | 0.6957 | 0.6426 | 0.0625 | 0.3526 |
0.5722 | 14.0 | 19908 | 0.7026 | 0.6390 | 0.0415 | 0.3402 |
0.5658 | 15.0 | 21330 | 0.6983 | 0.6432 | 0.0658 | 0.3545 |
0.5597 | 16.0 | 22752 | 0.7013 | 0.6405 | 0.0503 | 0.3454 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2