carbonated_sft_adapter

A LoRA/SFT adapter checkpoint published on Hugging Face by the Satya-Sri account, designed to apply parameter-efficient supervised fine-tuning (SFT) on top of a base language model for instruction-following or task-specific behavior. It is a small set of trained weights meant to be merged with or loaded alongside a compatible base model rather than used as a standalone model. Source: supabase.