File size: 1,324 Bytes
29b9eea 5a18ca8 29b9eea 5a18ca8 f0a1eb6 5a18ca8 f0a1eb6 c092f34 5a18ca8 f0a1eb6 5a18ca8 f0a1eb6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 |
---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: clinical-t5
results: []
datasets:
- AGBonnet/augmented-clinical-notes
language:
- en
metrics:
- rouge
pipeline_tag: summarization
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# clinical-t5
This is a finetuned T5-small model from Google, a checkpoint with 60 million parameters, for clinical note summarization.
It was finetuned with the [augmented-clinical-notes](https://huggingface.co/datasets/AGBonnet/augmented-clinical-notes) dataset, available in the Hugging Face.
## Intended uses & limitations
The model was created for learning purposes. Hence, although being briefly evaluated in [this](https://github.com/hossboll/clinical_nlp/blob/main/clinical_t5_finetuned.ipynb
) notebook, it should be further refined.
### 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: 4
### Framework versions
- Transformers 4.30.0
- Pytorch 2.2.1+cu121
- Datasets 2.19.1
- Tokenizers 0.13.3 |