Baidu Releases Unlimited OCR, a 3B Model That Keeps the KV Cache Flat for Long-Document Parsing
Baidu open-sourced Unlimited OCR, a 3B-parameter MoE model that parses dozens of document pages in a single forward pass. Its Reference Sliding Window Attention (R-SWA) holds the KV cache constant, so memory and latency stay flat as output grows. It scores 93.23 on OmniDocBench v1.5, beating the Dee...
Dense Supervision Is Not Enough: The Readout Blind Spot in Looped Language Models
arXiv:2606.24898v1 Announce Type: new
Abstract: Looped language models turn hidden states into runtime state: each state is decoded for prediction and fed back into future computation. This creates a basic supervision question: which state variables does cross-entropy actually control? We show that...
From Meta Idea to Advanced Mathematical Discovery -- Human-AI Co-Discovery of Sign-Embedding Quantum Algorithms
arXiv:2606.24899v1 Announce Type: new
Abstract: AI-assisted mathematics is often evaluated on solving predefined problems. In practice, however, many important advances begin earlier, when a vague research intuition is transformed into a concrete problem, a promising route, and a theorem family wor...
arXiv:2606.24900v1 Announce Type: new
Abstract: This paper proposes a new approach to near-sensor computing, in which a lightweight Neural Architecture Search (NAS) is performed directly on the deployment device to find the best tiny neural architecture for analyzing the real-time data acquired thr...
LLM Evolution as an Industry-Scale Ecosystem: A Lifecycle Perspective on Continual Learning
arXiv:2606.24901v1 Announce Type: new
Abstract: Continual learning capability is critical for Industrial LLMs, as deployed models must be continuously updated to meet evolving requirements and environments, rather than repeatedly retrained from scratch. However, most existing research focuses on im...
Project Auto-World: Towards Automated Benchmarking of Neural Relational Reasoners
arXiv:2606.24965v1 Announce Type: new
Abstract: Reasoning about relational structures remains a significant challenge for neural models, particularly when they must systematically apply learned knowledge to problem instances that are harder than those seen in training. Progress is hampered by the d...
Diagnosing and Mitigating Compounding Failures in Agentic Persuasion via Taxonomic Strategy Retrieval
arXiv:2606.24976v1 Announce Type: new
Abstract: Foundation-model agents in multi-step, open-ended environments frequently suffer from compounding errors, where early mistakes contaminate long-horizon trajectories. While Multi-Agent Debate (MAD) succeeds in deterministic domains, agents in subjectiv...
Do vision-language models search like humans? Reasoning tokens as a reaction-time analog in classic visual-search paradigms
arXiv:2606.25066v1 Announce Type: new
Abstract: Visual search has been one of the most productive paradigms in the study of visual attention: the way reaction time scales with the number of items distinguishes parallel, "pop-out" search from serial, attention-demanding search. I ask whether vision-...
Beyond Shapley: Efficient Computation of Asymmetric Shapley Values
arXiv:2606.25103v1 Announce Type: new
Abstract: We address the problem of explainability in machine learning models through feature attribution methods. In particular, we consider a variant of Shapley values known as Asymmetric Shapley Values (ASV), which enables the incorporation of causal knowled...
As ASML CEO Christophe Fouquet told TechCrunch in May, what China can currently buy are older-generation deep ultraviolet tools — gear first shipped about a decade ago — the same machines the MATCH Act would now put off limits.
This release brings the HFresh disk-based vector index and the built-in MCP Server to general availability, rebuilds cluster-wide async replication to run from a single scheduler (on by default), and adds two previews: the Boost API and Nested Object Filtering.
Gradium Launches stt-translate and s2s-translate, Real-Time Speech Translation Models Beating gpt-realtime-translate on Accuracy and Latency
Gradium released two real-time speech translation models, stt-translate and s2s-translate, covering English, French, German, Spanish, and Portuguese across 20 language pairs. The models collapse the standard three-model cascade into two, pairing single-pass transcription-and-translation with a Gradi...
Automated Schema Evolution in Pinterest’s Next-Generation DB Ingestion Framework
Yisheng Zhou | Software Engineer IILiang Mou | Sr Staff Software EngineerGabriel Raphael Garcia Montoya | Staff Software EngineerIstvan Podor | Staff Software EngineerIntroductionIn the first post of this series, we introduced Pinterest’s next-generation CDC-based ingestion platform built on Kafka, ...
AI Reality Check: The Hidden Cost of AI Adoption - Organizational Debt
Organizational debt—not compute or models—is the real barrier to AI ROI. Explore how process, cultural, and data debt quietly undermine enterprise AI adoption.
A Three-Phase Factual Recall Circuit in Gemma-2B and Gemma-12B-IT
Activation patching reveals how facts are stored, routed, and read out across transformer layers, and why the residual stream does most of the work
The post A Three-Phase Factual Recall Circuit in Gemma-2B and Gemma-12B-IT appeared first on Towards Data Science.
Talos: Scaling rare disease diagnosis with automated, iterative genomic reanalysis
Talos was built to help resolve a major bottleneck in genomic medicine: human review time. The open-source system recovered 90% of in-scope diagnoses while surfacing just 1.3 candidate variants per patient for expert review.
The post Talos: Scaling rare disease diagnosis with automated, iterative g...