|
|
--- |
|
|
base_model: |
|
|
- unsloth/Llama-3.2-3B-unsloth-bnb-4bit |
|
|
tags: |
|
|
- text-generation-inference |
|
|
- transformers |
|
|
- unsloth |
|
|
- llama |
|
|
- qlora |
|
|
license: apache-2.0 |
|
|
language: |
|
|
- en |
|
|
datasets: |
|
|
- miriad/miriad-4.4M |
|
|
pipeline_tag: text-generation |
|
|
library_name: transformers |
|
|
--- |
|
|
|
|
|
# π§ TibbScholar |
|
|
|
|
|
<img src="https://huggingface.co/Aasher/TibbScholar/resolve/main/TibbScholar_Banner.jpg" width="700"/> |
|
|
|
|
|
**TibbScholar** is a specialized, 3-billion-parameter language model fine-tuned for medical question-answering. It is based on Meta's `Llama-3.2-3B` model and is designed to serve as an informational tool for educational and research purposes. |
|
|
|
|
|
This model was trained on a large subset of the [MIRIAD-4.4M dataset](https://huggingface.co/datasets/miriad/miriad-4.4M) to provide concise, structured answers to medical queries. |
|
|
|
|
|
--- |
|
|
|
|
|
## π Model Details |
|
|
|
|
|
- **Base Model**: `unsloth/Llama-3.2-3B-unsloth-bnb-4bit` |
|
|
- **Fine-tuning Dataset**: A **100,000** record subset of [MIRIAD-4.4M](https://huggingface.co/datasets/miriad/miriad-4.4M). |
|
|
- **Prompt Format**: A simple Question/Answer structure (see below). |
|
|
- **Training Framework**: Fine-tuned using [Unsloth AI's](https://github.com/unslothai/unsloth) library with QLoRA for efficient training. |
|
|
|
|
|
--- |
|
|
|
|
|
## π Prompt Format |
|
|
|
|
|
For the model to perform as expected, prompts **must** follow the structure it was trained on. The prompt should end with `Answer:` followed by a newline. |
|
|
|
|
|
```text |
|
|
Question: |
|
|
What are the risks in dental implant surgery? |
|
|
|
|
|
Answer: |
|
|
``` |
|
|
--- |
|
|
|
|
|
## π‘ How to Use |
|
|
The model can be easily loaded using the transformers library. |
|
|
|
|
|
```python |
|
|
from transformers import pipeline |
|
|
import torch |
|
|
|
|
|
pipe = pipeline( |
|
|
"text-generation", |
|
|
model="Aasher/TibbScholar", |
|
|
torch_dtype=torch.bfloat16, # Or float16 for older GPUs |
|
|
device_map="auto", |
|
|
) |
|
|
|
|
|
prompt = """Question: |
|
|
What are the risks in dental implant surgery? |
|
|
|
|
|
Answer: |
|
|
""" |
|
|
|
|
|
response = pipe( |
|
|
prompt, |
|
|
max_new_tokens=256, |
|
|
do_sample=True, |
|
|
temperature=0.5, |
|
|
top_p=0.9, |
|
|
) |
|
|
|
|
|
print(response[0]["generated_text"]) |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
|
|
### β οΈ Intended Use and Limitations |
|
|
|
|
|
This model is intended solely for academic and informational purposes. It can be a helpful tool for students and researchers exploring medical topics. |
|
|
This model is NOT a medical professional. |
|
|
|
|
|
- The knowledge is limited to its training data and may not be up-to-date. |
|
|
- It can generate incorrect or incomplete information (hallucinate). |
|
|
- Do not use its outputs for clinical decision-making, diagnosis, or treatment. |
|
|
|
|
|
|
|
|
### π¨ Disclaimer: Not for Medical Advice |
|
|
The information provided by TibbScholar is not a substitute for professional medical advice. Always consult a qualified healthcare provider with any medical questions. The creator of this model assumes no liability for any actions taken based on its output. |
|
|
|
|
|
--- |
|
|
|
|
|
### π€ Acknowledgements |
|
|
|
|
|
- **Base Model:** Meta AI for the Llama 3.2 model. |
|
|
- **Dataset:** The creators of the MIRIAD dataset. |
|
|
- **Training:** The Unsloth AI team for their excellent fine-tuning library. |