HF Daily Papers · 2026-10-24

HF Daily Papers 精选（15 篇，按社区票数排序）

  • [222▲] WeVisOne-VL: Unified Visual Perception from Objects to Pixels — https://arxiv.org/abs/2610.31329
  • [164▲] Making MLLM Latents Matter: Factor-Aware Predictive Learning for Cinematic Video Generation — https://arxiv.org/abs/2610.30926
  • [152▲] Deep Search in Real-World Web Environments — https://arxiv.org/abs/2610.30181
  • [73▲] Editing Images into Reusable Layers via Compositional Edit-Prior Distillation — https://arxiv.org/abs/2610.24738
  • [54▲] IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems — https://arxiv.org/abs/2610.30949
  • [46▲] Toward Recursive Self-Improvement in Physical AI — https://arxiv.org/abs/2610.29677
  • [43▲] Policy-Mask: Budget-Aware Adaptation of Diffusion Features to Discriminative Tasks — https://arxiv.org/abs/2610.28889
  • [39▲] Deep Fusion Memory: A Chunk-Level Parametric Memory for Large Language Models — https://arxiv.org/abs/2610.25100
  • [32▲] What Generation Knows: Unlocking Fine-Grained Understanding in Unified Multimodal Models — https://arxiv.org/abs/2610.23987
  • [28▲] LaLa-World: Scaling Latent World Models with Natural Language Actions — https://arxiv.org/abs/2610.27851
  • [25▲] Meta Expert: Cross-Layer Expert Sharing for Storage- and Retrieval-Efficient MoE — https://arxiv.org/abs/2610.27598
  • [24▲] Code2Video-Compass: A Comprehensive Benchmark for General-Purpose Code-Based Video Generation across T2V, I2V, and V2V — https://arxiv.org/abs/2610.28416
  • [23▲] Self-Distilling Future Video Representations for World Action Models — https://arxiv.org/abs/2610.30320
  • [18▲] Heartbeat Thinking: Speculative Reasoning and Tool Execution While Listening in Realtime Voice Agents — https://arxiv.org/abs/2610.26233
  • [17▲] RouteWeaver: Weaving Mode Selection and Execution into Unified LLM Routing — https://arxiv.org/abs/2610.19705