This version focuses on explaining medical concepts, terminology, and common conditions, rather than conversation or patient-style dialogue (which appears in v2).

It responds to medical questions with:

Clear and structured explanations
General health guidance
Educational descriptions of symptoms, conditions, and anatomy
Safety reminders when symptoms may require professional care

🧠 Base Model

Base: mistralai/Mistral-7B-Instruct-v0.2
Finetuning: LoRA (QLoRA low-rank adaptation)
Framework: Unsloth + TRL + PEFT

This model does not perform diagnosis and does not provide medication dosing.

✅ Intended Use

Use Case	Allowed	Notes
Medical knowledge Q&A	✅	Education-focused
Explaining symptoms conceptually	✅	High-level guidance, no diagnosis
Anatomy, physiology, terminology explanations	✅	Good fit
Medical student study assistant	✅	Reliable structure, not clinical
Diagnosing medical conditions	❌	Must avoid guessing
Recommending specific treatments or doses	❌	High risk
Emergency care instructions	❌	Always refer to real professionals

⚖️ Safety Behavior

The model is trained to:

Avoid firm diagnoses
Encourage professional evaluation when appropriate
Not provide medication dosages
Avoid urgent-care decision instructions

This aligns the model with safe educational medical use.

🏋 Training Details

Dataset Used

Dataset	Size	Purpose
`keivalya/MedQuad-MedicalQnADataset`	~47k Q&A pairs	Medical education knowledge base

Data Preparation

Removed duplicates and malformed entries
Reformatted dataset into instruction–answer style
Ensured responses:
- Avoided prescriptive / treatment advice
- Emphasized explanation and general medical understanding

Training Configuration

Method: QLoRA Finetuning
Rank: 16
Precision: 4-bit
Sequence Length: 2048
Batching: Gradient Accumulation enabled for low-VRAM training

🧪 Example Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base = "mistralai/Mistral-7B-Instruct-v0.2"
adapter = "ysn-ir/mistral-medical-chat-lora-v1"

tokenizer = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(base, device_map="auto")
model = PeftModel.from_pretrained(model, adapter)

prompt = "What causes iron deficiency anemia?"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
output = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(output[0], skip_special_tokens=True))

🔔 Limitations

Not a medical professional
May simplify complex conditions
Not fit for emergency evaluation or treatment decisions
Should never be relied upon for real patient care

If the user reports:

Severe pain
Difficulty breathing
Loss of consciousness
Chest pain

→ The model must recommend immediate medical attention.

👤 Author

Hugging Face: https://huggingface.co/ysn-ir
Maintained by: ysn-ir

🧾 License

Released under Apache-2.0.
Permitted for:

Research
Personal use
Commercial use

Not permitted for clinical decision-making.

✨ Summary

This v1 model:

Focuses on medical Q&A and explanation
Good for education and understanding
Not intended for diagnosis or treatment

v2 builds on this with improved patient-style conversation safety behavior.

Use responsibly.