Skip to content
Only Research#Hugging Face
Nov 7Sat
  1. HuggingFace Daily Papers62

    WorldFoundry: Toward Unified Open-Source Infrastructure for World Intelligence

    WorldFoundry proposes open-source infrastructure for world intelligence, intended for reproducible inference and evaluation across video, 3D and 4D, embodied action, and interactive world systems. The work defines a generative world model as a predictive dynamical system that maintains a persistent, history-conditioned state representation, responds to actions or interventions, and generates temporally coherent trajectories. Its catalog includes 270 model entries, 74 benchmarks, and 62 metric entries, and it records execution readiness and evaluation conclusions separately through manifests, artifact contracts, and evidence-aware scorecards.

Nov 5Thu
  1. HuggingFace Daily Papers78

    PyRUA-DataFarm: Using LLM agents instead of teleoperation to generate real-ready robot data

    PyRUA-DataFarm proposes having a coding agent rebuild tasks in simulation, with a video model rendering them into real-ready data, replacing teleoperation for collecting robot demonstrations. On five real-robot tasks, policies trained only on this data achieved 93/100 successes, versus 84/100 for policies trained on teleoperated data; under an unseen tablecloth and colored lighting, the former kept 87 successes while the latter fell to 46.

    Why it matters: The paper presents a complete pipeline and cost comparison for using a coding agent to replace teleoperation in collecting demonstrations, which can be used to assess the room for replacing this data-collection route.

Nov 4Wed
  1. HuggingFace Daily Papers60

    Route Anything: Research on model and policy routing for general-purpose AI systems

    The study proposes Route Anything, a unified framework for studying model and policy routing in general-purpose AI systems, and establishes an optimal routing theory under quality-cost trade-offs. The authors construct an omni-modal trajectory matrix covering 25 benchmarks across language, image, video, speech, coding and workplace agents, and robotic manipulation, with 452 candidate configurations and 374,401 scoring runs.

  2. HuggingFace Daily Papers72

    DreamFree: A world action model without future video generation, reusing video diffusion priors via representation distillation

    DreamFree proposes a world action model (WAM) that does not require generating future videos. Through representation distillation, it turns a pretrained video diffusion model into a predictive visual encoder that predicts the frozen video model's representation using only the current observation; an action module then extracts action-relevant information.

Nov 1Sun
  1. HuggingFace Daily Papers72

    uni-robot-agent: A Unified Open-Source Framework for Robot-Use Agents

    A research team proposes uni-robot-agent, a unified framework to support further training of open-weight robot-use agents. Without privileged inputs, the out-of-the-box Qwen3.8-Flash-Next achieves a RoboDojo score of 31.39, surpassing the published GPT-6 Astra-RoboProbe result of 28.97, and performs competitively on 19 real-world tasks.

Oct 30Fri
  1. HuggingFace Daily Papers58

    Study: Probe evaluation of spatial state representations in visual foundation models

    The study uses frozen-feature probes to examine ten visual foundation model representations across four levels—spatial structure, object persistence, predictive spatial update, and actionability—under streaming first-person observations of static indoor environments. Results show that strong readout of the visible scene does not guarantee accurate object persistence or future-feature retrieval, and a spatial reader trained on real features does not transfer unchanged to predicted features. Object persistence and actionability depend on camera pose readout, while feature prediction depends on visible-scene readout.

Oct 29Thu
  1. HuggingFace Daily Papers79

    CaPtain: A Post-Training Pipeline for Code-as-Policy Robot Agents

    CaPtain proposes an open post-training pipeline that trains small open vision-language models such as Qwen3.5-4B into code-as-policy (CaP) robot agents that do not rely on VLAs, frontier models, or privileged state at test time.

    Why it matters: The work proposes using a physical critic to assign credit across multi-turn code policies and presents a complete training procedure for small models on LIBERO and real robots.

Oct 28Wed
Oct 26Mon
  1. arXiv cs.RO78

    Study: Full-weight VLA fine-tuning without an fp32 master copy silently erases most updates

    The paper points out that when VLA models are fully fine-tuned in bf16 without keeping an fp32 master copy, AdamW rounds away any update smaller than 1/512 of a weight; 9 of 12 open-source VLA codebases train this way by default or in documented recipes, including LeRobot's SmolVLA default configuration and the π0/π0.5 guides.

    Why it matters: The original quantifies the proportion of updates lost in bf16 fine-tuning and provides a LIBERO success-rate comparison for the fp32 master copy, which can serve as a reference for training configurations.

  2. HuggingFace Daily Papers59

    OSMA-Bench: An Obstacle-Conditioned Safety Benchmark for Household Robot Navigation

    The study proposes OSMA-Bench, an obstacle-conditioned safe motion adaptation benchmark built on spatially matched scenario sets that differ only in obstacle type, and introduces the Semantic Sensitivity Index (SSI) to measure whether a robot's clearance and its distance-dependent speed and smoothness follow the caution ordering of obstacles.

Oct 21Wed
  1. HuggingFace Daily Papers52

    Rethinking Video Storage for AI Training: Breaking Cross-Frame Dependencies for Efficient Random Access

    The paper identifies two types of decode amplification when randomly sampling videos for AI training: sequential reconstruction of unrelated frames, and reconstruction of the reference chains needed for sampled targets. The authors propose dependency-aware sparse decoding, which reconstructs only the requested frames and their necessary references without transcoding the source video; optional offline DDRA encoding constrains each interior frame between two intra-coded anchors, so that reconstruction of each requested target requires no more than three frames. On a large robotics corpus, the method achieves 1.45x the loading throughput of LeRobot v3 at similar reconstruction quality, with 58.7% less media data.

  2. HuggingFace Daily Papers58

    CausalAlign: Aligning Causal Representations of Video Generation Models for Stable Camera Ego-Motion

    CausalAlign proposes a causal alignment framework that uses linear probes and causal interventions to locate the representation layers for camera ego-motion in video generation models, then aligns the representations at those layers to a 3D foundation model via delta-token space. After fine-tuning Wan2.2 on DL3DV and RealEstate10K, geometric consistency improves by more than 20% over the base model, and on RoboCasa365 mobile manipulation tasks it raises the success rate of the world action model (WAM) by 10 percentage points.

  3. HuggingFace Daily Papers58

    TARSphere: A Full-Stack Framework and Benchmark for Long-Horizon Embodied Agents

    TARSphere proposes a full-stack framework for robots that evaluates long-horizon embodied agents under physical and social constraints. Its TARSphere-Infra runs robot policies in a physics simulator with embodied NPCs and constrained dialogue, and uses an event-driven finite-state machine to update task states based on completed actions and delivered information; TARSphere-Harness lets a frozen multimodal large language model revise its decisions through multi-level executable skills, observations, hierarchical memory, and execution feedback.

Oct 20Tue
Oct 12Mon
  1. 3D视觉工坊 · 微信公众号56

    Pumpire: A Unified Benchmark for Evaluating Point-to-Point Metric Distance Estimation in 3D Foundation Models

    Zhejiang University and other institutions propose Pumpire, a unified benchmark that directly evaluates the ability of image-level and video-level 3D foundation models to estimate point-to-point metric distances in reconstructed geometry, noting that previous approaches of separately evaluating depth and camera intrinsics cannot reflect this ability.