Unverified60% confidenceFactTime unknown
Self-Consistency通常将temperature设置为0.7以上以增加采样多样性
2
Sources
60%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
宁波银行AI Agent面试真题：推理优化+智能体调试+Python高频考点
bilibili互联网帮面试
Related Entities
Related Claims
Unverified自洽性采样的temperature通常设为0.7-1.079% similarUnverified自洽性做法是对同一问题进行多次高温采样（temperature通常设为0.7-1.0），再用多数投票选出频率最高的结果77% similarUnverified在测试场景中通常需要将Temperature调至较低值（如0.1-0.2）并配合固定随机种子以保证输出可复现性74% similarUnverified恒温精度标称±2℃，实际保温阶段温度波动在可接受范围，但对追求±1℃精准控温的用户来说不够精细72% similarUnverified较高的温度参数会增加输出的随机性，在测试场景中往往需要将其调低以保证输出稳定性71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/19829API
curl https://kongchang.com/api/v1/knowledge/claims/19829MCP
get_claim(id=19829)