Intelligence Digest

Top highlights (metadata-only; no AI synthesis):

- RESEARCH · Research: Mixed-State Long-Range Entanglement from Dimensional Constraints (https://arxiv.org/abs/2605.15201v1) — arXiv
snippet: We present a new mechanism for long-range entanglement (LRE) in strongly symmetric many-body mixed states that does not rely on symmetry anomalies or long-range correlations. Our primary example is the maximally mixed state in the transl…
- RESEARCH · Research: Translation symmetry-enforced long-range entanglement in mixed states (https://arxiv.org/abs/2605.15200v1) — arXiv
snippet: We show by a counting argument that even though translation symmetry admits symmetric short-range entangled (SRE) eigenstates, there are not enough such SRE eigenstates to span the zero momentum sector. This means that the fixed point st…
- RESEARCH · Research: EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generation (https://arxiv.org/abs/2605.15199v1) — arXiv
snippet: Multi-shot video generation extends single-shot generation to coherent visual narratives, yet maintaining consistent characters, objects, and locations across shots remains a challenge over long sequences. Existing evaluations typically…
- RESEARCH · Research: ATLAS: Agentic or Latent Visual Reasoning? One Word is Enough for Both (https://arxiv.org/abs/2605.15198v1) — arXiv
snippet: Visual reasoning, often interleaved with intermediate visual states, has emerged as a promising direction in the field. A straightforward approach is to directly generate images via unified models during reasoning, but this is computatio…
- RESEARCH · Research: RefDecoder: Enhancing Visual Generation with Conditional Video Decoding (https://arxiv.org/abs/2605.15196v1) — arXiv
snippet: Video generation powers a vast array of downstream applications. However, while the de facto standard, i.e., latent diffusion models, typically employ heavily conditioned denoising networks, their decoders often remain unconditional. We…
- RESEARCH · Research: VGGT-$Ω$ (https://arxiv.org/abs/2605.15195v1) — arXiv
snippet: Recent feed-forward reconstruction models, such as VGGT, have proven competitive with traditional optimization-based reconstructors while also providing geometry-aware features useful for other tasks. Here, we show that the quality of th…
- RESEARCH · Research: Non-Invertible Symmetries on Tensor-Product Hilbert Spaces and Quantum Cellular Automata (https://arxiv.org/abs/2605.15194v1) — arXiv
snippet: We investigate realizations of (1+1)-dimensional fusion category symmetries on tensor-product Hilbert spaces, allowing for mixing with quantum cellular automata (QCAs). It was argued recently that any such realizable symmetry must be wea…
- RESEARCH · Research: Aligning Latent Geometry for Spherical Flow Matching in Image Generation (https://arxiv.org/abs/2605.15193v1) — arXiv
snippet: Latent flow matching for image generation usually transports Gaussian noise to variational autoencoder latents along linear paths. Both endpoints, however, concentrate in thin spherical shells, and a Euclidean chord leaves those shells e…
- RESEARCH · Research: RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO (https://arxiv.org/abs/2605.15190v1) — arXiv
snippet: Causal autoregressive video diffusion models support real-time streaming generation by extrapolating future chunks from previously generated content. Distilling such generators from high-fidelity bidirectional teachers yields competitive…
- RESEARCH · Research: FutureSim: Replaying World Events to Evaluate Adaptive Agents (https://arxiv.org/abs/2605.15188v1) — arXiv
snippet: AI agents are being increasingly deployed in dynamic, open-ended environments that require adapting to new information as it arrives. To efficiently measure this capability for realistic use-cases, we propose building grounded simulation…
- RESEARCH · Research: Articraft: An Agentic System for Scalable Articulated 3D Asset Generation (https://arxiv.org/abs/2605.15187v1) — arXiv
snippet: A bottleneck in learning to understand articulated 3D objects is the lack of large and diverse datasets. In this paper, we propose to leverage large language models (LLMs) to close this gap and generate articulated assets at scale. We re…
- RESEARCH · Research: VGGT-Edit: Feed-forward Native 3D Scene Editing with Residual Field Prediction (https://arxiv.org/abs/2605.15186v1) — arXiv
snippet: High-quality 3D scene reconstruction has recently advanced toward generalizable feed-forward architectures, enabling the generation of complex environments in a single forward pass. However, despite their strong performance in static sce…