Articles & Guides

Hallo4D targets spatial and temporal hallucinations in 3D and 4D generation. Instead of retraining generators, it uses multimodal LLMs to detect inconsistencies and guide consensus-based corrections.

KnowAct-GUIClaw introduces a “Know Deeply, Act Perfectly” paradigm for personal GUI assistants, aiming to address OpenClaw’s limitations in cross-platform GUI interaction and self-evolution. The framework combines experience-based memory, a self-evolving skill library, and reflection to improve task execution over time.

GigaWorld-Policy-0.5 tackles a key bottleneck in World Action Models: costly video generation during inference. It keeps future visual dynamics as a training signal while switching to action-only decoding for deployment.

A Hugging Face Blog post argues that newer, broader OCR models can still lose to a narrowly specialized system. On a Brazilian Portuguese benchmark, DharmaOCR outperformed Mistral OCR4 and Unlimited-OCR by focusing its training capacity on one language domain.

Hugging Face disclosed a security incident in which an autonomous AI agent system allegedly carried out a multi-stage intrusion against part of its production infrastructure. The case highlights how data-processing pipelines and model platforms are becoming first-class security targets.

VideoRAE explores whether frozen video foundation model representations can become compact, reconstructable, and generation-friendly video latents. The work targets a key limitation of conventional 3D-VAEs: latents optimized mainly for pixel reconstruction.

A new arXiv paper introduces MOJO, a joint training framework that combines masked autoencoding with supervised decoding for spike-tokenizing neural models. The approach aims to make better use of unlabelled neural recordings, especially when behavioural labels are scarce.

A new arXiv paper proposes measuring non-Gaussianity with the squared Wasserstein distance to a standard Gaussian. The resulting OT-ICA algorithm aims to replace common proxy contrasts used in classical ICA.

A new arXiv paper reframes the race toward generative game engines around a classic loop: action, state, and observation. Its central message is that playable worlds need persistent rules and evolving states, not just convincing video prediction.

MetaPerch explores a simple but underused idea: bioacoustic models should learn not only from animal sounds, but also from the metadata attached to those recordings. The work uses signals such as location and time as auxiliary supervision for more robust species identification.

A new arXiv paper tests whether frozen Evo 2 representations contain linearly accessible biosecurity-relevant information. Lightweight probes show strong performance on antimicrobial resistance detection, while virulence and generated-sequence labels remain more limited.

Ai2’s Shippy shows that production agents in high-stakes domains are not just model wrappers. Reliability comes from tool design, sandboxing, explicit limits, and workflow-level evaluation.

IBM Research argues that model routing in agentic systems is less a classification problem than a systems optimization problem. Real-world performance depends on cost, latency, caching, infrastructure, and governance constraints.

The Hindcast paper proposes a time-aware evaluation setup for LLM forecasters, replaying resolved Polymarket questions as if models were standing at an earlier date. Its goal is to separate genuine forecasting from answer leakage through retrieval or training data.

A new arXiv paper introduces Deep Interaction, a human-in-the-loop method that lets users edit faulty reasoning steps in an LLM’s original answer and reuse the corrected chain to steer future output.

A new arXiv paper introduces Earthquaker-AI, a hybrid framework that combines Lego WeDo2 robotics, a retrieval-augmented conversational assistant, and rubric-based assessment for earthquake preparedness education.

A new arXiv paper proposes an AI-accelerated framework for professional upskilling that spans knowledge acquisition, content creation, review, teaching, and assessment. Its main contribution is treating workforce learning as an integrated production-and-learning pipeline rather than a set of isolated AI tools.

A new arXiv paper studies a multi-expert OCR pipeline for historical Manchu documents with sharply different visual styles. The system routes each page to a specialist model using a lightweight page-level image classifier.

A new arXiv paper introduces PAT, a RAG-based system designed to push LLM translation beyond sentence-by-sentence rendering. The study finds that specifications and corpus-informed context can encourage broader reformulation, though the benefits remain inconsistent.

A new arXiv study of 2,361 popular GitHub repositories finds that agentic coding tools are entering open-source workflows, but intensive use remains concentrated in a small number of projects.

A new arXiv paper examines feature selection in renewable power prediction and proposes Cluster-based Sequential Feature Selection. CSFS matches the predictive performance of standard sequential feature selection while cutting computational cost by 21% on average.