Unverified50% confidenceFactExact time
Decide Live's scoring formula is P(option_i) = softmax_i(19.96 * cos(e_query, e_option_i)), where the scaling factor 19.96 is learned and corresponds to a temperature of approximately 0.05.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe UP host uses Unsloth's recommended sampling parameters: temperature of 1, top K of 20, repeat penalty of 1, and min P of 0.65% similarVerifiedThe Softmax formula with temperature parameter T is P(xᵢ) = exp(zᵢ/T) / Σexp(zⱼ/T); when T→0 it degenerates to argmax, and when T→∞ it approaches a uniform distribution65% similarUnverified温度参数（Temperature）的数学本质是对Softmax函数输出的概率分布进行缩放，趋近于0时模型进行贪心解码65% similarUnverified在LLM层面微量随机因子对应调高采样温度（Temperature）或Top-P参数，工程做法是检测到轨迹重复时将Temperature从默认值（如0.2）动态提升至0.7~1.064% similarVerifiedtemperature为0时退化为贪婪解码，每步选取概率最高的Token，但仅能保证单次模型版本内的确定性62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002777API
curl https://kongchang.com/api/v1/knowledge/claims/1002777MCP
get_claim(id=1002777)