Yinghui He

Skill-switching accuracy in LLMs drops significantly on complex tasks, but a new training approach boosts performance from 34.4% to 68.4% on challenging benchmarks.

Targeted middle-layer recurrence can dramatically enhance Transformer reasoning without the need for full-layer looping, leading to superior performance in complex tasks.

Privileged self-distillation can paradoxically hinder thinking models, leading to a 17% drop in accuracy on long reasoning tasks due to its impact on learning dynamics.

Prioritizing domains based on their cross-domain transferability can boost multi-domain RLVR performance by up to 10%.