Unverified50% confidenceDecision RuleExact time
When the tail of the weight distribution is very heavy, fold-wise targeting is more stable; when the weights are mild, pooled targeting has smaller variance because it uses all the data, and there is no significant difference between the two.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedA larger smoothness parameter β indicates a smoother density function, theoretically enabling better estimation accuracy with fewer samples.73% similarVerifiedThe 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.73% similarUnverifiedRMSNorm相比LayerNorm对数值范围约束更强，输出分布更集中，减少量化时的溢出风险71% similarUnverified优质Prompt在大多数场景下比微调更高效，在数据量不足或需求模糊的阶段贸然做微调往往事倍功半70% similarUnverified温度超参数τ越小分布越集中，模型对难负样本越敏感，但τ过小会导致梯度对异常值敏感、训练不稳定70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023535API
curl https://kongchang.com/api/v1/knowledge/claims/1023535MCP
get_claim(id=1023535)