Medical LLM fine-tuned on 14+ medical datasets. Achieves 68.2% accuracy across USMLE, AIIMS, and NEET medical benchmarks — a 4.42% improvement over base Mixtral.
| Benchmark | Accuracy |
|---|---|
| Combined (USMLE/AIIMS/NEET) | 68.2% |
| Improvement over base | +4.42% |
Training: LoRA (rank 64, alpha 16), LR 2e-4, 3 epochs across 14+ medical datasets including MedQA, MedMCQA, Healthcare Magic 100k.
Available on Hugging Face.
Hosted on Hugging Face.