AdaExploreBench evaluates whether LLM agents can actively discover hidden environmental rules and adapt to unannounced ...

MIST separates scene setup from change execution to isolate memory failures in video world models, and M3-Agent improves ...

Remote KV caching for long-context LLM serving can severely degrade latency unless the store is sharded by parallel rank, ...

OneWM unifies causal training of world and action modeling, showing that learning action-conditioned dynamics improves ...

Velox accelerates long video rollouts by routing attention via geometric frustum slabs, reducing latency while preserving ...

W e p l a c e d f o u r G e m i n i 4 A r g o n a g e n t s i n a s a n d b o x e d M i n e c r a f t s u r v i v a l w o r l ...

The method converts pre-trained transformers into elastic models that adapt continuously to varying computational budgets by ...

Streaming vision-language models (VLMs) must continuously integrate information from the past while remaining responsive to ...

Multi-teacher on-policy distillation (MOPD) is used in two settings. In common-domain composition, several teachers score ...

Index-Translate-35B-A3B-preview is the 35B-total / 3B-active mixture-of-experts multilingual text model of the Index-Translate family, built on Qwen3.5. It supports text translation across 150 ...