edbeeching/fineproofs-qwen3-235b-a22b-thinking-2507
收藏资源简介:
FineProofs Qwen3 235B A22B Thinking 2507是一个与FineProofs兼容的训练数据集快照,使用Qwen/Qwen3-235B-A22B-Thinking-2507模型生成。该快照包含4202行数据,生成于2026年5月27日。数据集的schema与lm-provers/FineProofs-SFT完全一致,包括问题、推理内容、证明、类别、竞赛、gemini-3-pro-grade、qwen3-4b-thinking-reward@128、来源和消息等列。数据集仅包含完整的生成结果(finish_reason为stop),且推理和证明内容非空;grade/reward列存在但值为空,因为这些生成的证明尚未被评分。
FineProofs Qwen3 235B A22B Thinking 2507 is a FineProofs-compatible training snapshot generated with the Qwen/Qwen3-235B-A22B-Thinking-2507 model. The snapshot contains 4202 rows and was generated on May 27, 2026. The schema exactly matches lm-provers/FineProofs-SFT, including columns such as problem, reasoning_content, proof, category, competition, gemini-3-pro-grade, qwen3-4b-thinking-reward@128, source, and messages. Only complete generations with finish_reason=stop and non-empty reasoning and proof are included; the grade/reward columns are present for schema compatibility but left null because these generated proofs have not been graded.




