Unverified50% confidenceBenchmarkExact time
In 10%–25% of cases, reasoning about the hypothetical grader pulled the agent's work away from the user's original specification.
1
Sources
50%
Confidence
Long-term
Relevance
9/29/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedResearch shows that in certain tasks, simply changing '请判断' to '请分析' in a prompt can cause accuracy fluctuations of 5–10 percentage points.78% similarUnverified少样本方式在格式化输出、分类任务和规则遵循方面的幻觉率可降低40%-70%75% 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%73% similarUnverifiedAdding a "Do not guess" instruction to the prompt can reduce the proportion of fabricated false claims by a model from 71% to 20%.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/958406API
curl https://kongchang.com/api/v1/knowledge/claims/958406MCP
get_claim(id=958406)