111 lines
2.0 KiB
Markdown
111 lines
2.0 KiB
Markdown
|
---
|
||
|
|
library_name: transformers
|
||
|
|
tags:
|
||
|
|
- qwen
|
||
|
|
- biomedical
|
||
|
|
- bioinformatics
|
||
|
|
- fine-tuned
|
||
|
|
- medical
|
||
|
|
- llm
|
||
|
|
license: apache-2.0
|
||
|
|
base_model:
|
||
|
|
- Qwen/Qwen2.5-1.5B
|
||
|
|
---
|
||
|
|
|
||
|
|
# Qwen2.5-1.5B Biomedical Fine-Tuned Model
|
||
|
|
|
||
|
|
This model is a biomedical and bioinformatics fine-tuned version of **Qwen/Qwen2.5-1.5B**, fine-tuned by **Dr. YMG**.
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Model Details
|
||
|
|
|
||
|
|
### Model Description
|
||
|
|
|
||
|
|
This model is a domain-adapted and instruction fine-tuned large language model specialized for biomedical and bioinformatics tasks.
|
||
|
|
|
||
|
|
- Developed by: Dr. YMG
|
||
|
|
- Model type: Causal Language Model (LLM)
|
||
|
|
- Language(s): English
|
||
|
|
- License: Apache 2.0
|
||
|
|
- Finetuned from model: Qwen/Qwen2.5-1.5B
|
||
|
|
|
||
|
|
### Model Sources
|
||
|
|
|
||
|
|
- Repository: https://huggingface.co/yashm/qwen25-15b-biomed-finetuned
|
||
|
|
- Base Model: https://huggingface.co/Qwen/Qwen2.5-1.5B
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Uses
|
||
|
|
|
||
|
|
### Direct Use
|
||
|
|
|
||
|
|
- Biomedical concept explanation
|
||
|
|
- Bioinformatics discussions
|
||
|
|
- Research assistance
|
||
|
|
- Literature summarization
|
||
|
|
- Gene expression & biomarker discussion
|
||
|
|
|
||
|
|
### Out-of-Scope Use
|
||
|
|
|
||
|
|
- Clinical diagnosis
|
||
|
|
- Medical treatment decisions
|
||
|
|
- Drug prescription
|
||
|
|
- Patient-specific advice
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Example Usage
|
||
|
|
|
||
|
|
```python
|
||
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
|
||
|
|
import torch
|
||
|
|
|
||
|
|
MODEL_ID = "yashm/qwen25-15b-biomed-finetuned"
|
||
|
|
|
||
|
|
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
|
||
|
|
model = AutoModelForCausalLM.from_pretrained(
|
||
|
|
MODEL_ID,
|
||
|
|
device_map="auto",
|
||
|
|
dtype=torch.bfloat16,
|
||
|
|
trust_remote_code=True,
|
||
|
|
)
|
||
|
|
|
||
|
|
pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)
|
||
|
|
|
||
|
|
prompt = "Explain gene expression in simple terms."
|
||
|
|
|
||
|
|
out = pipe(prompt, max_new_tokens=200)
|
||
|
|
print(out[0]["generated_text"])
|
||
|
|
```
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Training Details
|
||
|
|
|
||
|
|
- Base model: Qwen/Qwen2.5-1.5B
|
||
|
|
- Method: LoRA (PEFT)
|
||
|
|
- Precision: BF16
|
||
|
|
- Quantization: 4-bit QLoRA
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Limitations
|
||
|
|
|
||
|
|
- May hallucinate
|
||
|
|
- Not medically validated
|
||
|
|
- Limited to training data
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Disclaimer
|
||
|
|
|
||
|
|
For research and educational use only. Not for clinical decision-making.
|
||
|
|
|
||
|
|
---
|
||
|
|
|
||
|
|
## Author
|
||
|
|
|
||
|
|
Fine-tuned by Dr. YMG
|