Datasetpublic

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).

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face

0likes23downloads

Dataset Card

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, sourcefile, type, question, answer, context, llmjudgescore, llmjudgeverdict, llmjudge_reason. Used to fine-tune Sudhanshu1985/slm-125m-sft. Types: lookup / reasoning / unanswerable (refusals).