Unverified50% confidenceFactExact time
Post-distillation preference optimization can improve the student's harm-avoidance ability, but cannot eliminate the differences brought by the source teacher.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedExisting few-step distillation methods fall roughly into two categories: one relies on an additional score model to estimate the teacher distribution, and the other introduces adversarial losses to bring the student's output distribution closer to the teacher's.65% similarUnverifiedCompared to adjusting reward weights, constrained reinforcement learning offers two advantages: communicability and composability, as adding a new constraint does not force the recalibration of all weights.64% similarUnverifiedStrategies for mitigating reward design problems include inverse reinforcement learning, intrinsic motivation methods, and constrained reinforcement learning.63% similarUnverifiedReinforcement learning in the post-training stage can improve answer accuracy on both in-distribution and out-of-distribution tasks, but does not consistently improve the validity of reasoning traces.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023612API
curl https://kongchang.com/api/v1/knowledge/claims/1023612MCP
get_claim(id=1023612)