nli_sts_roberta_large_lr1e-05_wd1e-03_ep10_lr1e-05_wd1e-03_ep10_ckpt
This model is a fine-tuned version of ys7yoo/sts_roberta_large_lr1e-05_wd1e-03_ep10 on the klue dataset. It achieves the following results on the evaluation set:
- Loss: 0.6903
- Accuracy: 0.8963
- F1: 0.8962
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: 1e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
---|---|---|---|---|---|
0.6445 | 1.0 | 391 | 0.4254 | 0.852 | 0.8512 |
0.2943 | 2.0 | 782 | 0.3371 | 0.889 | 0.8886 |
0.1586 | 3.0 | 1173 | 0.3704 | 0.888 | 0.8881 |
0.0921 | 4.0 | 1564 | 0.4429 | 0.892 | 0.8919 |
0.0565 | 5.0 | 1955 | 0.4864 | 0.899 | 0.8989 |
0.0378 | 6.0 | 2346 | 0.5727 | 0.8963 | 0.8962 |
0.0238 | 7.0 | 2737 | 0.6247 | 0.8957 | 0.8955 |
0.016 | 8.0 | 3128 | 0.6578 | 0.8947 | 0.8945 |
0.0101 | 9.0 | 3519 | 0.6780 | 0.8953 | 0.8952 |
0.0067 | 10.0 | 3910 | 0.6903 | 0.8963 | 0.8962 |
Framework versions
- Transformers 4.33.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.0
- Tokenizers 0.13.3
- Downloads last month
- 78
Finetuned from
Dataset used to train ys7yoo/nli_sts_roberta_large_lr1e-05_wd1e-03_ep10_lr1e-05_wd1e-03_ep10_ckpt
Evaluation results
- Accuracy on kluevalidation set self-reported0.896
- F1 on kluevalidation set self-reported0.896