aarajbhattarai/rejected-nepali-law-v2
Nepali Source-Grounded Instruction Dataset — REJECTED Synthetic Nepali instruction-tuning data generated with NVIDIA NeMo Data Designer from authoritative Nepali documents (agriculture manuals, legal texts). Answers are grounded strictly in the source; unanswerable questions get an explicit refusal. Records use chat messages format plus metadata and per-record quality_scores (grounding / correctness / naturalness, 1-5, LLM-as-judge). One data/train-<shard>.jsonl per source… See the full description on the dataset page: https://huggingface.co/datasets/aarajbhattarai/rejected-nepali-law-v2.
Nepali Source-Grounded Instruction Dataset — REJECTED
Synthetic Nepali instruction-tuning data generated with NVIDIA NeMo Data Designer from authoritative Nepali documents (agriculture manuals, legal texts). Answers are grounded strictly in the source; unanswerable questions get an explicit refusal. Records use chat messages format plus metadata and per-record quality_scores (grounding / correctness / naturalness, 1-5, LLM-as-judge). One data/train-<shard>.jsonl per source document; shards are overwritten idempotently on re-runs.
Records judged BELOW the quality gate; each has reject_reasons. Useful for judge calibration, hard-negative mining, or re-filtering with different thresholds. Do NOT use as-is for instruction tuning.