Unverified50% confidenceFactExact time
When a model encounters out-of-distribution inputs, prediction entropy increases, the softmax output tends toward a uniform distribution, and the range of candidate tokens permitted by nucleus sampling expands, giving long-tail anomalous tokens a chance of being sampled.
1
Sources
50%
Confidence
Long-term
Relevance
8/14/2026
First Seen
Sources
Related Claims
UnverifiedSoftmax输出的概率值并不等同于真实的贝叶斯后验概率，而是被过度优化以拉开类别间距，导致预测概率系统性偏高72% similarUnverifiedSoftmax函数将神经网络logits压缩为概率分布，但其输出值在大规模模型中系统性偏离真实贝叶斯后验概率72% similarUnverifiedCross-entropy loss often converges faster than MSE in classification tasks because, when paired with softmax output, gradients do not vanish in the saturation region.70% similarUnverifiedThe higher the temperature parameter, the flatter the probability distribution output by Softmax, and the more creative and diverse the output; the lower the temperature, the sharper the distribution and the more deterministic and conservative the output.66% similarUnverified权重初始化的随机性对模型性能影响最深远，数据批次打乱顺序的随机性影响最小66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/747862API
curl https://kongchang.com/api/v1/knowledge/claims/747862MCP
get_claim(id=747862)