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 ...
The method converts pre-trained transformers into elastic models that adapt continuously to varying computational budgets by ...
Excited to share ATAS — Any Time, Any Space! Video models can produce impressive individual shots, but keeping a whole story consistent is harder: a room should keep its layout when revisited, and ...
Join the discussion on this paper page LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation ...
How can we build an agent that thinks beyond human language, yet still acts through it in the external environment? Modern LLM agents still reason and interact through text, while ...
Join the discussion on this paper page Player-1: Learning to Play from Synthetic Experience ...
Join the discussion on this paper page FlashVSR+: Efficient Video Super-Resolution via Hybrid Attention and Reward-Yield Distribution Matching ...
Join the discussion on this paper page Are We Testing Too Much? A Survey of Benchmark Compression for Efficient Evaluation of Large Language Models ...
Join the discussion on this paper page LifeSearch: Benchmarking Deep Research Agents for Real-World Life and Productivity ...
Join the discussion on this paper page Published on Oct 13 ·Submitted by Jiahao Meng on Oct 14 ·Peking University Upvote 19 +11 Authors: Jiahao Meng, Zhongyuan Bao , Yue Tan , Haiquan Wen , Renrui ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results