metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- squad
base_model: FabianWillner/bert-base-uncased-finetuned-triviaqa
model-index:
- name: bert-base-uncased-finetuned-triviaqa-finetuned-squad
results: []
bert-base-uncased-finetuned-triviaqa-finetuned-squad
This model is a fine-tuned version of FabianWillner/bert-base-uncased-finetuned-triviaqa on the squad dataset. It achieves the following results on the evaluation set:
- Loss: 0.9981
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: 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: 2
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.0184 | 1.0 | 5533 | 0.9733 |
| 0.7496 | 2.0 | 11066 | 0.9981 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0+cu113
- Datasets 2.3.2
- Tokenizers 0.12.1