llama-1b-multimedqa-c75b25-kcenter-raw

A 1B-parameter fine-tuned Llama checkpoint targeting medical or multi-domain question answering, produced via a k-center data selection recipe on the MultiMedQA dataset. It is a task-specific derivative of Meta's Llama 3.2 1B base model intended for low-resource or local deployment. Source: supabase.