platzi-distilroberta-base-mrpc-glue-joselier
This model is a fine-tuned version of distilroberta-base on the glue and the mrpc datasets. It achieves the following results on the evaluation set:
- Loss: 0.5993
- Accuracy: 0.8211
- F1: 0.8809
Model description
This model uses transfer learning to classify 2 sentences (a string of 2 sentences separated by a comma) in "Equivalent" or "Not Equivalent". The model platzi-distilroberta-base-mrpc-glue-joselier was programmed as part of a class from Platzi's course "Curso de Transfer Learning con Hugging Face"
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: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- 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 | Accuracy | F1 |
|---|---|---|---|---|---|
| 0.5365 | 1.09 | 500 | 0.5993 | 0.8211 | 0.8809 |
| 0.3458 | 2.18 | 1000 | 0.8336 | 0.8235 | 0.8767 |
Framework versions
- Transformers 4.32.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3