This dataset contains Romanized Nepali Alpaca-like data converted to ShareGPT format.
Each entry represents a conversation including instruction, question, and context (if present).
The "human" turn includes an instruction (if present), context (if present), and the question.
The "gpt" turn contains the response.
Suitable for fine-tuning large language models for Romanized Nepali instruction following and RAG.… See the full description on the dataset page:
https://huggingface.co/datasets/dineshkarki/roman-nepali-alpaca-sharegpt.
