Search papers, labs, and topics across Lattice.
3
0
6
0
Adaptive self-distillation can boost model performance by over 23 points without extra rollouts, reshaping how we approach teacher-student dynamics in training.
DeltaFlow-P achieves a remarkable 2.72x speedup in denoising while reducing perplexity, challenging the dominance of full attention in language modeling.
Stop letting simulator errors in critical regions derail your policies: Sim2Act aligns surrogate fidelity with downstream decision impact, leading to more stable and robust decision-making.
