distilbert-base-uncased-finetuned-squad
This model is a fine-tuned version of distilbert-base-uncased on the squad dataset. It achieves the following results on the evaluation set:
- Loss: 1.2104
Model description
Most base model weights were frozen leaving only to finetune the last layer (qa outputs) and 3 last layers of the encoder.
Training and evaluation data
Achieved EM: 73.519394512772, F1: 82.71779517079237
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: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.3937 | 1.0 | 5533 | 1.2915 |
| 1.1522 | 2.0 | 11066 | 1.2227 |
| 1.0055 | 3.0 | 16599 | 1.2104 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.17.0
- Tokenizers 0.10.3
- Downloads last month
- 14
Inference Providers
NEW
This model is not currently available via any of the supported third-party Inference Providers, and
the model is not deployed on the HF Inference API.