Unverified50% confidenceFactExact time
In knowledge distillation, the temperature coefficient T typically ranges from 3 to 20; the higher the temperature, the smoother the probability distribution.
1
Sources
50%
Confidence
Long-term
Relevance
9/7/2026
First Seen
Sources
Related Entities
Related Claims
Unverified采样温度T=1时保持原始概率分布，T→0时输出趋于确定性，T>1时分布更平坦、输出更随机79% similarVerified温度（temperature）参数控制采样过程，温度越高输出越随机多样，温度越低越趋向最高概率的选择74% similarVerifiedThe Temperature parameter of large models typically ranges from 0 to 2; the lower the temperature, the more deterministic and conservative the output, while the higher the temperature, the more diverse and creative the output.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/869707API
curl https://kongchang.com/api/v1/knowledge/claims/869707MCP
get_claim(id=869707)