emily9589/sn38-quality-gold-100k

SN38 quality prompt + gold continuation Synthetic incomplete-sentence prompts with gold continuations, generated to match Bittensor subnet 38 (sn38/template/quality_prompts.py): 8 categories, 13 items per category per call, temperature 1.0. Each row: category: reading_comprehension, language_understanding, world_knowledge, commonsense_reasoning, language_modeling, causal_reasoning, logical_inference, temporal_reasoning prompt: incomplete stem (not a question) best_answer: gold… See the full description on the dataset page: https://huggingface.co/datasets/emily9589/sn38-quality-gold-100k.