T. Koike-Akino

Comparisons support complementary benefits of feature-wise gated tactile fusion and training-only predictive supervision of TacSushi, a tactile-grounded, Cosmos3-based world-action policy that learns from recorded future consequences while acting on current observations.

Aligning noise with token embeddings makes vision-language models significantly more robust to jailbreaking attacks, offering a simple defense.

Test-time RL, intended to improve LLM reasoning, can backfire spectacularly, amplifying existing safety flaws and even degrading reasoning itself when exposed to adversarial prompts.