Sudhanshu1985/slm-125m-qa-dataset
slm-125m QA dataset (SFT) 24,713 reviewed grounded-QA pairs (LLM-judged, kept >=4) over US case law, SEC filings, and educational web text. Columns: source, source_file, type, question, answer, context, llm_judge_score, llm_judge_verdict, llm_judge_reason. Used to fine-tune Sudhanshu1985/slm-125m-sft. Types: lookup / reasoning / unanswerable (refusals).
024
Upload qna_combined_final.csv with huggingface_hub
Upload README.md with huggingface_hub
initial commit
