albert-base-v2-finetuned-squad
This model is a fine-tuned version of albert-base-v2 on the squad dataset. It achieves the following results on the evaluation set:
- Loss: 1.4539
- Exact Match: 80.60548722800378
- F1 score: 88.76870326468953
Model description
This model is fine-tuned on the extractive question answering task -- The Stanford Question Answering Dataset -- SQuAD2.0.
Intended uses & limitations
More information needed
Training and evaluation data
Training and evaluation was done on SQuAD2.0.
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 |
|---|---|---|---|
| 0.8702 | 1.0 | 5540 | 0.8943 |
| 0.6972 | 2.0 | 11080 | 0.9087 |
| 0.4998 | 3.0 | 16620 | 0.9890 |
| 0.3601 | 4.0 | 22160 | 1.1892 |
| 0.235 | 5.0 | 27700 | 1.4539 |
Framework versions
- Transformers 4.34.0
- Pytorch 1.12.1
- Datasets 2.14.5
- Tokenizers 0.14.1