Search papers, labs, and topics across Lattice.
Chinese Academy of Sciences ♣
3
0
7
UCOB achieves unprecedented performance in agentic reinforcement learning by dynamically refining skill usage through credit-aware self-distillation.
LLMs train 1.5x faster and generalize better with a surprisingly simple trick: adapt learning rates per-layer based on the "heavy-tailedness" of their weight matrices.
Self-play can be dramatically improved by exploiting the "question construction path" it generates as privileged information for self-distillation, leading to 2-3x faster learning.
