Unverified60% confidenceFactTime unknown
机器学习领域约有30%-50%的论文没有公开可用的代码
1
Sources
60%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedResearch indicates that more than 60% of machine learning paper results are difficult to independently reproduce.68% similarUnverifiedApproximately 23% of papers could not be fully reproduced as claimed, and at least 496 papers had at least one claim that was falsified or disputed.61% similarUnverified2024 年的多项基准测试表明，经过严格安全对齐的模型在代码生成任务上的拒绝率比基座模型高出 15-30%，在创意写作任务上的自我审查率高达 40%60% similarUnverified实验数据显示，经过5-10轮迭代训练后，模型在低频词汇和长尾知识上的覆盖率下降超过40%，而输出文本在表面流畅性上几乎没有可察觉的退化60% similarVerifiedGoogle's engineering practice research indicates that code review can detect approximately 60% of defects, far exceeding the approximately 30% detection rate of unit testing.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/29467API
curl https://kongchang.com/api/v1/knowledge/claims/29467MCP
get_claim(id=29467)