Search papers, labs, and topics across Lattice.
7
0
13
5
Small visual perturbations can cause World-Action Models to execute harmful actions while still predicting a plausible future, revealing a critical vulnerability in their design.
dOPSD leverages a model's own decoding process to provide on-policy supervision, leading to substantial gains in reasoning performance without external labels.
Flex-Forcing achieves superior video generation quality and stability by unifying autoregressive and bidirectional methods, all while speeding up inference.
dMoE slashes the memory footprint of Mixture-of-Experts Diffusion LLMs by up to 80% without sacrificing performance, finally making them practical.
Naively quantizing autoregressive video diffusion models tanks performance due to exponentially increasing error accumulation across frames and heterogeneous outlier patterns, but Q-ARVD solves it.
Forget hand-crafted membership inference attacks - AutoMIA learns better strategies automatically, adapting to different models and eliminating the need for manual feature engineering.
Linear attention models can now achieve SOTA controllable generation performance, thanks to a new gated conditioning module that overcomes the limitations of ControlNet-style approaches.
