Unverified50% confidenceSolutionExact time
DPO preference optimization can, on top of SFT, further align output quality using preference pairs of good and poor headnotes.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedCISPO（Clipped Importance Sampling Policy Optimization）与 PPO 同源，通过重要性采样比值的裁剪约束策略更新幅度68% similarUnverified在监督微调（SFT）中，数据的分布往往比数据的数量更关键，精准的数据补充胜过盲目扩充数据集63% similarUnverified将LLM通过率纳入质量门禁涉及阈值设定、执行效率优化和回归检测三个关键决策62% similarUnverifiedTop-p (nucleus sampling) limits the candidate token range by dynamically truncating probability mass, and has better adaptability than fixed-truncation Top-k62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/910242API
curl https://kongchang.com/api/v1/knowledge/claims/910242MCP
get_claim(id=910242)