dmis-lab/llama-3-meerkat-70b-v1.0

dmis-lab/llama-3-meerkat-70b-v1.0 is a 70 billion parameter instruction-tuned medical AI system from the Meerkat model family, developed by dmis-lab. Based on Meta's Llama-3-70B-Instruct, it is fine-tuned on a synthetic dataset of chain-of-thought reasoning paths from 18 medical textbooks and diverse instruction-following datasets. This model excels in high-level medical reasoning and problem-solving, achieving an average of 77.9% across seven medical benchmarks with an 8192-token context length.