Downloads · 30 days
80
9% of all-time downloads
ahmedalkhair64/qwen2.5-14b-medical-sft
qwen2.5-14b-medical-sft is a machine learning model from ahmedalkhair64. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
A medical reasoning fine-tune of Qwen2.5-14B-Instruct using LoRA (rank 64) trained on the English subset of the FreedomIntelligence/medical-o1-reasoning-SFT dataset via LLaMA-Factory.
Downloads · 30 days
80
9% of all-time downloads
All-time downloads
938
Public
Parameters
14.8B
38.5 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors29.5 GB · 77%
From the Hugging Face model README
A medical reasoning fine-tune of Qwen2.5-14B-Instruct using LoRA (rank 64) trained on the English subset of the FreedomIntelligence/medical-o1-reasoning-SFT dataset via LLaMA-Factory.
This model is designed to improve medical reasoning capabilities, particularly for complex clinical scenarios requiring multi-step diagnostic reasoning. It was trained on medical case studies with detailed reasoning chains covering various specialties including cardiology, neurology, pulmonology, endocrinology, and infectious diseases.


This model works best with llama.cpp using the quantized GGUF format:
Important: Use temperature 0.3 for accurate medical reasoning. Higher temperatures cause hallucinations.
This model does NOT work with ollama. The GGUF Q4_K_M quantization format used here is different from ollama's internal quantization system. Attempting to use this model with ollama will result in errors or poor performance.
Solution: Use llama.cpp instead (see Usage section above).
Always use temperature 0.3 for medical applications.
Trained on English medical data only. Performance on other languages is not guaranteed.
Recommended context length: 4096 tokens. Some complex medical cases may require longer context.
Apache 2.0 (inherited from base model and dataset)