Research Papers
ID-Forcing Tackles Drift in Long Autoregressive Video Generation
A new test-time framework, In-Distribution Forcing (ID-Forcing), keeps cached key-value entries in-distribution to stop color, tex...
Hugging Face Paper: LoHi Reframes Long-Video Efficiency Around Frames, Pixels, and Decoding
A new Hugging Face paper argues that long-video understanding should trade per-frame resolution for denser temporal coverage, and...
OSWorld-Pro Introduces Process-Based Evaluation for Computer-Use Agents
A new benchmark of over 300 tasks and 2,800 subgoals, grounded in 67,000 human annotations, evaluates computer-use agents on their...
CANOPY Brings Adaptive-Granularity Compression to Multimodal RAG
Hugging Face paper CANOPY proposes a post-retrieval compression framework that represents retrieved text, tables, images and video...
ASCENT Lets Deployed LLM Agents Learn From Their Own Verified Trajectories
A new method called ASCENT enables long-horizon LLM agents to keep improving during deployment by self-distilling verified experie...
Kandinsky 6.0 Video Debuts Dual-Stream Models for Synchronized Audio-Video Generation
Hugging Face paper introduces Kandinsky 6.0 Video, a family of diffusion foundation models in 3B Lite and 29B Pro sizes that gener...
ALoDLM: Hugging Face Paper Introduces Token-Adaptive Looped Diffusion Language Models
A new paper proposes ALoDLM, which replaces the uniform computation used by diffusion language models with token-adaptive latent r...
Hugging Face Paper Proposes SMI: Using MLLMs to Manage Spatial Memory in World Models
A new paper introduces Spatial Memory Intelligence (SMI), a framework that uses a multimodal understanding model to manage spatial...
EditHero: New Benchmark Tests Long-Horizon, Part-Level 3D Editing
Hugging Face researchers introduce EditHero, a benchmark for long-horizon, part-level 3D editing with natural-language instruction...
PDE-JEPA: Predictive Representation Learning for Parametric PDE Dynamics
Hugging Face paper introduces PDE-JEPA, a predictive pretraining framework for parametric PDEs that combines masked-latent predict...
Hugging Face Paper Benchmarks 'Scientific Slop' in AI-Generated Research
A new preprint introduces SciSlopBench, a 390-paper benchmark for detecting 'scientific slop' — plausible-looking AI papers whose...
LexReward Brings Taxonomy-Driven Reward Modeling to Legal AI
A new Hugging Face paper proposes LexReward, a framework that scores legal responses along Style, Element, and Chain dimensions us...