Knowledge catalogue
arXiv:2605.11387v1 Announce Type: new Abstract: We address the problem of fine-tuning pre-trained generative policies with reinforcement learning (RL) while preserving the multimodality of their actio
arXiv:2605.12483v1 Announce Type: new Abstract: In settings where labeled verifiable training data is the binding constraint, each checked example should be allocated carefully. The standard practice
arXiv:2605.12413v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) show strong visual perception, yet remain limited in reasoning about space under changing viewpoints. We study
arXiv:2605.11221v1 Announce Type: cross Abstract: Predictive models in biomedicine depend on structured assay data locked in the text, tables, and supplements of primary publications. This bottleneck
arXiv:2605.11438v1 Announce Type: new Abstract: Medical imaging research has spent a decade getting very good at one thing: producing per-voxel masks. Masks tell us size, volume, and location, and a d
arXiv:2605.11857v1 Announce Type: new Abstract: Federated fine-tuning of large language models is commonly formulated as a parameter aggregation problem. However, even parameter-efficient methods requ
arXiv:2501.16931v2 Announce Type: replace Abstract: Machine learning models are often evaluated using point estimates of performance metrics such as accuracy, F1 score, or mean squared error. Such sum
arXiv:2605.11904v1 Announce Type: new Abstract: The Nearest Class Mean (NCM) classifier is widely favored in Class-Incremental Learning (CIL) for its superior resistance to catastrophic forgetting com
arXiv:2605.11460v1 Announce Type: new Abstract: System identification (SysID) is critical for modeling dynamical systems from experimental data, yet traditional approaches often fail to capture nonlin
arXiv:2605.11287v1 Announce Type: new Abstract: A persistent paradox in time-series forecasting is that structurally simple MLP and linear models often outperform high-capacity Transformers. We argue
arXiv:2605.12271v1 Announce Type: new Abstract: Humans often specify and create through visual artifacts: typography sheets, sketches, reference images, and annotated scenes. Yet modern visual generat
This X post from Harrison Chase likely discusses anticipated features or developments in a major product update or release, inviting community feedback on what users are most excited about. The post a
arXiv:2605.11829v1 Announce Type: cross Abstract: The accurate recovery of constituent-level optical properties from integrating sphere measurements is a central analytical challenge in pharmaceutical
arXiv:2605.11107v1 Announce Type: new Abstract: Vision-language models (VLMs), such as CLIP and SigLIP 2, are widely used for image classification, yet their vision encoders remain vulnerable to syste
arXiv:2605.11577v1 Announce Type: new Abstract: Autoregressive language models generate text one token at a time, yet natural language is inherently structured in multi-token units, including phrases,
arXiv:2602.00767v2 Announce Type: replace Abstract: Emergent misalignment can arise when a language model is fine-tuned on a narrowly scoped supervised objective: the model learns the target behavior,
arXiv:2605.11726v1 Announce Type: new Abstract: Recently, reinforcement learning (RL) has been widely applied during post-training for diffusion large language models (dLLMs) to enhance reasoning with
Blue Origin CEO Dave Limp told employees that the company needs external capital to significantly increase launch frequency, stating that achieving established launch goals requires 'a large amount of
Security operations platform startup Blumira Inc. today launched the pilot of Kindling, an agentic security information and event management investigation engine that the company says can reduce alert
我們開源了這顆星球🌎上速度最快的低成本 bm25 引擎。 so we built psql_bm25s. exact BM25 retrieval. native Postgres access method. ~23x faster than pg_search on the standard benchmark. retrieval stops being a budget item. the
Somewhere inside every large company, there is a long list of AI pilots — promising, applauded, then quietly shelved. The culprit, more often than not, is a data activation failure. The gap between AI
The AI arms race has a new front, and it isn’t the battle over individual models — it’s war for the entire AI control plane. That theme was front and center at Boomi World in Chicago, where Boomi LP C
arXiv:2512.12131v2 Announce Type: replace Abstract: The scale of transformer model pre-training is constrained by the increasing computation and communication cost. Low-rank bottleneck architectures o
arXiv:2605.12034v1 Announce Type: cross Abstract: Omni-modal language models are intended to jointly understand audio, visual inputs, and language, but benchmark gains can be inflated when visual evid
This episode features Emad Mostaque, founder of Stability AI, in a live podcast discussion hosted by David Johnston. The conversation likely covers topics related to AI development, Stability AI's pro
arXiv:2508.07642v3 Announce Type: replace-cross Abstract: Vision-and-Language Navigation (VLN) poses significant challenges for agents to interpret natural language instructions and navigate complex 3
arXiv:2605.11461v1 Announce Type: cross Abstract: Reinforcement learning with verifiers (RLVR) has become a central paradigm for improving LLM reasoning, yet popular group-based optimization algorithm
arXiv:2605.11748v1 Announce Type: new Abstract: Bronchoscopy is routinely conducted in pulmonary clinics and intensive care units, but navigating the complex branching of the respiratory tract remains
arXiv:2605.12339v1 Announce Type: new Abstract: Aligning language models for both helpfulness and safety typically requires complex pipelines-separate reward and cost models, online reinforcement lear
buckle up because cinema is leaping ahead. I just finished watching a number of truly excellent short AI films in the @runwayml AI film festival, and I am honestly blown away by the level of quality.
This post demonstrates how to build a documentation extraction and model fine-tuning pipeline that addresses challenges when processing the complex financial documents. By combining Pulse AI's advance
Building end-to-end live streaming applications with real-time voice interaction presents several challenges. This post introduces a solution based on Amazon Nova 2 Sonic (Nova Sonic) and Amazon Kines
This article describes OpenAI's approach to creating a secure sandbox environment that safely executes code generated by Codex on Windows systems. It likely covers the technical challenges of isolatin
A Chrome extension that communicates directly with local Ollama instances without sending data to external servers , enabling users to submit messages through a popup interface with responses streamin
Built a local coding harness powered by Gemma 4. It runs locally, connects to my model backend, starts coding sessions, streams responses, and uses tools through a CLI-style workflow. Still early, but
arXiv:2605.11723v1 Announce Type: new Abstract: In this paper, we propose Concentrate and Concentrate (CaC), a coarse-to-fine anomaly reward model based on Vision-Language Models. During inference, it
arXiv:2605.12266v1 Announce Type: new Abstract: Graph-based machine learning has emerged as a promising approach for manufacturability analysis by learning directly from CAD models represented as Boun
arXiv:2511.12034v2 Announce Type: replace Abstract: Multimodal representation learning harmonizes distinct modalities by aligning them into a unified latent space. Recent research generalizes traditio
arXiv:2605.11442v1 Announce Type: cross Abstract: Large Language Model (LLM) agents have emerged as key intermediaries, orchestrating complex interactions between human users and a wide range of digit
arXiv:2605.11300v1 Announce Type: new Abstract: Vision state space models inherit the efficiency and long-range modeling ability of Mamba-style selective scans. However, their performance depends crit
arXiv:2604.03061v2 Announce Type: replace Abstract: Recent advances in generative AI raise the question of whether general-purpose image editing models can serve as unified solutions for image restora
arXiv:2605.12028v1 Announce Type: new Abstract: We describe our system for SemEval-2026 Task 8 (MTRAGEval), participating in Task A (Retrieval) across four English-language domains. Our approach emplo
arXiv:2605.11705v1 Announce Type: new Abstract: The training of large multimodal models fundamentally relies on massive image-text datasets, which inevitably incur prohibitive computational overhead.
arXiv:2605.11186v1 Announce Type: new Abstract: Auto-regressive decoding in Large Language Models (LLMs) is inherently memory-bound: every generation step requires loading the model weights and interm
arXiv:2605.11373v1 Announce Type: cross Abstract: The trustworthiness of AI decision-making systems is increasingly important. A key feature of such systems is the ability to provide recommendations f
arXiv:2605.11365v1 Announce Type: cross Abstract: Automated systems built on artificial intelligence (AI) are increasingly deployed across high-stakes domains, raising critical concerns about fairness
arXiv:2605.11362v1 Announce Type: new Abstract: In the data-driven era, large-scale datasets are routinely collected and analyzed using machine learning (ML) and artificial intelligence (AI) to inform
arXiv:2605.12496v1 Announce Type: new Abstract: Autoregressive video generation aims at real-time, open-ended synthesis. Yet, cinematic storytelling is not merely the endless extension of a single sce
Process mining software company Celonis SE said today it has snapped up the Massachusetts Institute of Technology-linked decision intelligence startup Ikigai Labs Inc., planning to use it to power a n
Cerebras, a company specializing in AI accelerators and wafer-scale computing systems, is discussed in terms of its approaches to improving token generation speed in large language models, which is cr
arXiv:2602.09368v2 Announce Type: replace Abstract: Gradient-based methods can efficiently optimize controllers by leveraging differentiable simulation and physical priors. However, contact-rich manip
arXiv:2505.05665v4 Announce Type: replace-cross Abstract: Large language models (LLMs) have recently demonstrated success in decision-making tasks including planning, control, and prediction, but thei
The term 'human slop' refers to low-effort content created by humans, sometimes harder to detect than AI-generated content because it fails to expose itself immediately. The concept emerged as a count
arXiv:2605.11533v1 Announce Type: new Abstract: Clinical check-up reports are multimodal documents that combine page layouts, tables, numerical biomarkers, abnormality flags, imaging findings, and dom
arXiv:2605.11304v1 Announce Type: new Abstract: Chest radiograph interpretation requires temporal reasoning over prior and current studies, yet most vision-language models are trained on static image-
William Sandlund / Financial Times: China's CSI AI Index, which tracks AI supply chain companies, is up 28%+ YTD, while HK's Hang Seng Tech Index, which includes Alibaba and Tencent, is down 8% — Tenc
Owen Walker / Financial Times: China's move to block Meta's Manus acquisition challenges the “Singapore washing” model, as major Chinese tech companies build significant Singapore presences — Beijing'
arXiv:2605.11779v1 Announce Type: new Abstract: Multiword expressions (MWEs) are a heterogeneous set with a glaring need for classifications. Designing a satisfactory classification involves choosing
arXiv:2605.11960v1 Announce Type: new Abstract: Vision Large Language Models (VLLMs) have achieved remarkable success in modern text-rich visual understanding. However, their perceptual robustness in
arXiv:2605.11335v1 Announce Type: cross Abstract: Layerwise offloading reduces the GPU memory footprint of large diffusion transformer (DiT) inference by prefetching upcoming layers from host memory,