Downloads · 30 days
17
3% of all-time downloads
longluu/Medical-QA-gatortrons-COVID-QA
Medical-QA-gatortrons-COVID-QA is a question answering model from longluu. Use it when the input is a question plus a passage. It is set up for transformers. The card lists the license as mit.
The model is an extractive Question Answering algorithm that can find an answer to a question by finding a segment in a text.
Downloads · 30 days
17
3% of all-time downloads
All-time downloads
578
Public
Parameters
354M
4.3 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors1.4 GB · 100%
From the Hugging Face model README
The model is an extractive Question Answering algorithm that can find an answer to a question by finding a segment in a text.
The base pretrained model is GatorTronS which was trained on billions of words in various clinical texts (https://huggingface.co/UFNLP/gatortronS). Then using the COVID-QA dataset (https://huggingface.co/datasets/covid_qa_deepset), I fine-tuned the model for an extractive Question Answering algorithm that can answer a question by finding it within a text.
The github code associated with the model can be found here: https://github.com/longluu/Medical-QA-extractive.
This dataset contains 2,019 question/answer pairs annotated by volunteer biomedical experts on scientific articles regarding COVID-19 and other medical issues. The dataset can be found here: https://github.com/deepset-ai/COVID-QA. The preprocessed data can be found here https://huggingface.co/datasets/covid_qa_deepset.
The hyperparameters are --per_device_train_batch_size 4
--learning_rate 3e-5
--num_train_epochs 2
--max_seq_length 512
--doc_stride 250
--max_answer_length 200 \
The model was trained and validated on train and validation sets.
Here we use 2 metrics for QA tasks exact match and F-1.
{'exact_match': 37.12871287128713, 'f1': 64.90491019877854}
Feel free to reach out to me at [email protected] if you have any question or suggestion.