Base model: Llama2-7B

Trained on MedMCQA + PubMedQA

Full training

Acc: 0.45