opt-small-sw
This model is a fine-tuned version of facebook/opt-125m on the SwitchBoard dataset. It achieves the following results on the evaluation set:
- Loss: 2.7166
- Accuracy: 0.4663
Model description
More details soon.
Intended uses & limitations
More details soon.
Training and evaluation data
More details soon.
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 7
- eval_batch_size: 7
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20.0
Training results
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1
- Datasets 2.9.0
- Tokenizers 0.13.0.dev0