hBERTv2_stsb
This model is a fine-tuned version of gokuls/bert_12_layer_model_v2 on the GLUE STSB dataset. It achieves the following results on the evaluation set:
- Loss: 0.9534
- Pearson: 0.7722
- Spearmanr: 0.7707
- Combined Score: 0.7714
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 | Pearson | Spearmanr | Combined Score |
|---|---|---|---|---|---|---|
| 4.4386 | 1.0 | 23 | 2.5331 | 0.1313 | 0.1071 | 0.1192 |
| 1.8741 | 2.0 | 46 | 2.0517 | 0.4923 | 0.4766 | 0.4844 |
| 1.347 | 3.0 | 69 | 1.3556 | 0.6964 | 0.7079 | 0.7022 |
| 0.8443 | 4.0 | 92 | 1.2583 | 0.7340 | 0.7367 | 0.7353 |
| 0.5822 | 5.0 | 115 | 0.9534 | 0.7722 | 0.7707 | 0.7714 |
| 0.4356 | 6.0 | 138 | 1.1921 | 0.7798 | 0.7771 | 0.7785 |
| 0.3531 | 7.0 | 161 | 1.3849 | 0.7701 | 0.7700 | 0.7700 |
| 0.2712 | 8.0 | 184 | 1.0015 | 0.7886 | 0.7870 | 0.7878 |
| 0.259 | 9.0 | 207 | 1.0523 | 0.7898 | 0.7874 | 0.7886 |
| 0.2003 | 10.0 | 230 | 1.1525 | 0.7836 | 0.7824 | 0.7830 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.14.0a0+410ce96
- Datasets 2.10.1
- Tokenizers 0.13.2
- Downloads last month
- 2
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Dataset used to train gokuls/hBERTv2_stsb
Evaluation results
- Spearmanr on GLUE STSBvalidation set self-reported0.771