111 lines
3.8 KiB
Markdown
111 lines
3.8 KiB
Markdown
|
---
|
|||
|
|
license: apache-2.0
|
|||
|
|
datasets:
|
|||
|
|
- mteb/raw_medrxiv
|
|||
|
|
language:
|
|||
|
|
- en
|
|||
|
|
- zh
|
|||
|
|
base_model:
|
|||
|
|
- prithivMLmods/Qwen3-1.7B-ft-bf16
|
|||
|
|
pipeline_tag: text-generation
|
|||
|
|
library_name: transformers
|
|||
|
|
tags:
|
|||
|
|
- trl
|
|||
|
|
- text-generation-inference
|
|||
|
|
- medical
|
|||
|
|
- article
|
|||
|
|
- biology
|
|||
|
|
- med
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
# **Canum-med-Qwen3-Reasoning (Experimental)**
|
|||
|
|
|
|||
|
|
> **Canum-med-Qwen3-Reasoning** is an **experimental medical reasoning and advisory model** fine-tuned on **Qwen/Qwen3-1.7B** using the **MTEB/raw\_medrxiv** dataset.
|
|||
|
|
> It is designed to support **clinical reasoning, biomedical understanding, and structured advisory outputs**, making it a useful tool for researchers, educators, and medical professionals in experimental workflows.
|
|||
|
|
|
|||
|
|
> \[!note]
|
|||
|
|
> GGUF: [https://huggingface.co/prithivMLmods/Canum-med-Qwen3-Reasoning-GGUF](https://huggingface.co/prithivMLmods/Canum-med-Qwen3-Reasoning-GGUF)
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## **Key Features**
|
|||
|
|
|
|||
|
|
1. **Medical Reasoning Focus**
|
|||
|
|
Fine-tuned on **MTEB/raw\_medrxiv**, enabling strong performance in **biomedical literature understanding**, diagnostic reasoning, and structured medical advisory tasks.
|
|||
|
|
|
|||
|
|
2. **Clinical Knowledge Extraction**
|
|||
|
|
Summarizes, interprets, and explains medical research papers, case studies, and treatment comparisons.
|
|||
|
|
|
|||
|
|
3. **Step-by-Step Advisory**
|
|||
|
|
Provides structured reasoning chains for **symptom analysis, medical explanations, and advisory workflows**.
|
|||
|
|
|
|||
|
|
4. **Evidence-Aware Responses**
|
|||
|
|
Optimized for scientific precision and evidence-driven output, suitable for **research assistance** and **medical tutoring**.
|
|||
|
|
|
|||
|
|
5. **Structured Output Mastery**
|
|||
|
|
Capable of producing results in **LaTeX**, **Markdown**, **JSON**, and **tabular formats**, supporting integration into research and healthcare informatics systems.
|
|||
|
|
|
|||
|
|
6. **Optimized for Mid-Scale Deployment**
|
|||
|
|
Balanced efficiency for **research clusters**, **academic labs**, and **edge deployments in healthcare AI prototypes**.
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## **Quickstart with Transformers**
|
|||
|
|
|
|||
|
|
```python
|
|||
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|||
|
|
|
|||
|
|
model_name = "prithivMLmods/Canum-med-Qwen3-Reasoning"
|
|||
|
|
|
|||
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|||
|
|
model_name,
|
|||
|
|
torch_dtype="auto",
|
|||
|
|
device_map="auto"
|
|||
|
|
)
|
|||
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|||
|
|
|
|||
|
|
prompt = "Summarize the findings of a study on the effectiveness of mRNA vaccines for COVID-19."
|
|||
|
|
|
|||
|
|
messages = [
|
|||
|
|
{"role": "system", "content": "You are a medical reasoning assistant that explains biomedical studies and provides structured clinical insights."},
|
|||
|
|
{"role": "user", "content": prompt}
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
text = tokenizer.apply_chat_template(
|
|||
|
|
messages,
|
|||
|
|
tokenize=False,
|
|||
|
|
add_generation_prompt=True
|
|||
|
|
)
|
|||
|
|
|
|||
|
|
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
|||
|
|
|
|||
|
|
generated_ids = model.generate(
|
|||
|
|
**model_inputs,
|
|||
|
|
max_new_tokens=512
|
|||
|
|
)
|
|||
|
|
generated_ids = [
|
|||
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
|||
|
|
print(response)
|
|||
|
|
```
|
|||
|
|
|
|||
|
|
---
|
|||
|
|
|
|||
|
|
## **Intended Use**
|
|||
|
|
|
|||
|
|
* **Medical research summarization** and literature review
|
|||
|
|
* **Diagnostic reasoning assistance** for educational or research purposes
|
|||
|
|
* **Clinical advisory explanations** in structured step-by-step format
|
|||
|
|
* **Biomedical tutoring** for students and researchers
|
|||
|
|
* **Integration into experimental healthcare AI pipelines**
|
|||
|
|
|
|||
|
|
## **Limitations**
|
|||
|
|
|
|||
|
|
* ⚠️ **Not a replacement for medical professionals** – should not be used for direct clinical decision-making
|
|||
|
|
* Training limited to research text corpora – may not capture rare or real-world patient-specific contexts
|
|||
|
|
* Context length limits restrict multi-document medical record analysis
|
|||
|
|
* Optimized for reasoning and structure, not empathetic or conversational dialogue
|