mbart-large-50-Biomedical_Dataset
This model is a fine-tuned version of facebook/mbart-large-50.
It achieves the following results on the evaluation set:
- Training Loss: 1.0165
- Epoch: 1.0
- Step: 2636
- Validation Loss: 0.9425
- Bleu: 38.9893
- Rouge Metrics:
- Rouge1: 0.6826259612196924
- Rouge2: 0.473675987811788
- RougeL: 0.6586445010303293
- RougeLsum: 0.6585487473231793
- Meteor: 0.6299677745833094
- Prediction lengths: 24.362727392855568
Model description
For more information on how it was created, check out the following link: https://github.com/DunnBC22/NLP_Projects/blob/main/Machine%20Translation/Biomedical%20Translation%20(EN%20to%20IT)/Biomedical%20-%20Translation%20Project.ipynb
Intended uses & limitations
This model is intended to demonstrate my ability to solve a complex problem using technology.
Training and evaluation data
Dataset Source: https://huggingface.co/datasets/paolo-ruggirello/biomedical-dataset
Histogram of English Input Word Counts
/Images/Histogram%20of%20English%20Lengths.png)
Histogram of Italian Input Word Counts
/Images/Histogram%20of%20Italian%20Inputs.png)
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
Training results*
| Training Loss | Epoch | Step | Validation Loss | Bleu | Rouge1 | Rouge2 | RougeL | RougeLsum | Meteor | Prediction Lengths |
|---|---|---|---|---|---|---|---|---|---|---|
| 1.0165 | 1.0 | 2636 | 0.9425 | 38.9893 | 0.6826 | 0.4737 | 0.6586 | 0.6585 | 0.6270 | 24.3627 |
Footnotes:
*: All results in this table are rounded to the nearest ten-thousandths of the decimal.
Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1
- Datasets 2.13.1
- Tokenizers 0.13.3
