roberta-base-finetuned-squad_v2
This model is a fine-tuned version of roberta-base on the SQuAD2.0. It achieves the following results on the evaluation set:
- Loss: 1.2340
- Exact Match: 79.684
- F1-score: 83.159
Overview
Language model: roberta-base
Language: English
Downstream-task: Extractive QA
Training data: SQuAD 2.0
Eval data: SQuAD 2.0
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.8978 | 1.0 | 8239 | 0.8532 |
0.6621 | 2.0 | 16478 | 0.9733 |
0.5045 | 3.0 | 24717 | 0.9010 |
0.3837 | 4.0 | 32956 | 1.0523 |
0.3069 | 5.0 | 41195 | 1.2340 |
Framework versions
- Transformers 4.34.0
- Pytorch 1.12.1
- Datasets 2.14.5
- Tokenizers 0.14.1
- Downloads last month
- 9
Finetuned from
Dataset used to train lauraparra28/Roberta-base-finetuned-SQuAD2.0
Evaluation results
- Exact Match on squad_v2validation set self-reported83.159
- F1 on squad_v2validation set self-reported79.684