Knowledge catalogue
// The Efficiency Frontier in LLMs // (bookmark this one) How much are you overpaying for context you do not need? It turns out that context costs dominate production LLM bills, and the right strategy
The Frenchman (@ylecun) seems incapable of giving credit where credit is due. @GaryMarcus is correct about LeCun's lack of integrity. Here he is rephrasing something that everyone else in the business
Adi Robertson / The Verge: The FTC settles with Cox, MindSift, and 1010 Digital Works for $930K over claims they falsely said they could use phone mics to spy on users for ad targeting — More specific
The Grok Build team is one of the most productive teams I have worked with. I need something to adopt my diverse workflow from model training to validating the Cybercab manufacturing process. Their da
arXiv:2605.23087v1 Announce Type: new Abstract: Neural collapse (NC) describes the structured geometry that emerges in the features and weights of trained classifiers. Recent theory suggests NC can be
I spent five years inside Microsoft Corp.’s Windows business during an era when Dell Technologies Inc. was a top-volume partner. These two companies sold in lockstep. Every Dell personal computer ship
arXiv:2605.22842v1 Announce Type: cross Abstract: Multi-agent AI pipelines typically assume that agent misconduct originates from model misalignment. We identify a structural failure in this assumptio
arXiv:2605.23778v1 Announce Type: cross Abstract: Could it be that AI weather models are solving physical equations, although they may not be the equations used by conventional NWP models? We compute
The Pope rightly warns that AI must serve human dignity, not become a tool of domination or exclusion. But if we hand governments sweeping power over AI development in the name of safety, how do we pr
The ramp up of cancer immunotherapy is remarkable. Now we're seeing vaccines achieve some cures or remissions in the most refractory cancers: pancreatic, melanoma, glioblastoma, renal, triple-negative
arXiv:2605.22870v1 Announce Type: cross Abstract: Chain-of-thought (CoT) prompting is necessary for arithmetic in small language models, yet shuffling its steps preserves most performance. What does C
arXiv:2601.21306v2 Announce Type: replace-cross Abstract: This paper investigates search in model-based reinforcement learning (RL). Conventional wisdom holds that long-term predictions and compoundin
The team is gathering speed Grok Build just got another update: v0.1.220-alpha.1 xAI is now shipping multiple updates in a single day with nonstop fixes, improvements, optimizations, and new features
the tech bros think they know better than me they think they know better than @michaeljburry they think they know better than the Nobelist @DAcemogluMIT they think they know better than @sundarpichai
arXiv:2605.23045v1 Announce Type: cross Abstract: Video representation learning has seen tremendous progress in recent years. This has been driven by many factors, including the scale of training and
Charles Clover / Financial Times: The UK, outpaced by the US and China in AI, is turning to experimental technologies like neuromorphic computing in search of computing sovereignty — Military capabili
the @xai team really cooked with Grok Build the most beautiful (and performant) TUI I've seen so far >realtime context view >feels snappy in ghostty >first class worktree support >great resume session
Thinking of AI as a productivity booster for prior workflows is the wrong framing. Like all of the previous waves of computerization/softwarization, AI is a tool that lets you do new things in new way
this bit on physics is deep and actually resonates with the fact that LLMs are equally comfortable learning any world, whereas humans are built for our world. See also Chomsky’s 2023 conversation with
This creates problems for the @Kevinroose narrative that AGI is near. Finally, a big name has the courage to tell it: we are nowhere near AGI. Demis Hassabis, CEO of Google DeepMind and Nobel laureate
This trend will continue for many years SpaceX put 10 megawatts of solar power in space across 3000 gen1 Starlink satellites, then they put 100 megawatts in space with 7000 gen2. soon, they're doing 1
arXiv:2605.23081v1 Announce Type: new Abstract: Efficient attention algorithms are critical to mitigate the quadratic cost of attention in long-context workloads. Prior work utilises block-scaled quan
arXiv:2506.04390v2 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) systems are vulnerable to attacks that inject poisoned passages into the retrieved context, even at low c
time to get out of index funds? they will soon be saddled with the IPOs of three giant companies that barely make a profit (or don’t make profits at all). I don’t personally want to be part of that. (
Today, I'm working on trying to better understand how coding assistants mention HF's products. Taking a simple approach of running tons of queries and analyzing the answers with @DAKlingbeil's https:/
arXiv:2605.22853v1 Announce Type: cross Abstract: Many modern datasets are large and carry complex structural relationships. Graph-based methods have traditionally been used to represent networked dat
arXiv:2605.23872v1 Announce Type: new Abstract: We introduce training-free looped transformers, in which a lightweight inference-time wrapper loops a contiguous mid-stack block of layers of a frozen c
arXiv:2603.04005v2 Announce Type: replace-cross Abstract: The rate-distortion-perception (RDP) tradeoff characterizes the fundamental limits of lossy compression by jointly considering bitrate, recons
arXiv:2605.22902v1 Announce Type: cross Abstract: Generative Vision-Language Models (VLMs) perform well on multimodal reasoning, but how visual inputs are transformed to text remains poorly understood
arXiv:2603.01655v2 Announce Type: replace Abstract: Ray tracing has become a standard for accurate radio propagation modeling, but suffers from exponential computational complexity, as the number of c
arXiv:2602.19174v5 Announce Type: replace Abstract: Natural language processing for the Turkic language family, spoken by over 200 million people across Eurasia, remains fragmented, with most language
arXiv:2605.23257v1 Announce Type: cross Abstract: Navigating under non-stationary environment shifts poses a critical challenge for a Vision-and-Language Navigation (VLN) agent deployed in the wild. Y
arXiv:2605.23274v1 Announce Type: new Abstract: Retrieving events from large-scale video datasets is challenging due to complex temporal, spatial, and multimodal information. This paper presents U-CES
arXiv:2605.23098v1 Announce Type: new Abstract: Reliable uncertainty estimation is critical for deploying monocular depth deep neural networks (DNNs) in safety-critical robotic systems. Conventional u
arXiv:2605.22968v1 Announce Type: cross Abstract: Machine learning methods provide a methodological innovation that can help screen for cardiovascular disease through noninvasive and readily available
arXiv:2605.23033v1 Announce Type: cross Abstract: Foundational Models pretrained on huge amount of data learn representations that evolve across depth, forming a hierarchy of embeddings with distinct
arXiv:2605.23171v1 Announce Type: cross Abstract: Recent advancements in instructional fine-tuning have injected noise into embeddings, with NEFTune (Jain et al., 2024) setting benchmarks using unifor
arXiv:2605.23565v1 Announce Type: cross Abstract: Reinforcement learning agents often exhibit unintended goal-directed behaviour outside their training distribution, but we currently lack a principled
arXiv:2603.18123v3 Announce Type: replace-cross Abstract: Foundation models promise to unify multiple clinical tasks within a single framework, but recent ultrasound studies report that unified models
arXiv:2605.23464v1 Announce Type: new Abstract: We consider a decentralized setup in which the participants collaboratively train and serve a large neural network, and where each participant only proc
arXiv:2507.23372v2 Announce Type: replace Abstract: Emotional understanding and generation are often treated as separate tasks, yet they are inherently complementary and can mutually enhance each othe
arXiv:2503.12868v2 Announce Type: replace Abstract: Learning-based medical image registration has matched the accuracy of conventional methods while offering superior computational efficiency. However
arXiv:2408.03085v3 Announce Type: replace-cross Abstract: As the most central and computationally intensive component of deep neural networks, the execution efficiency of matrix multiplication directl
arXiv:2510.07869v4 Announce Type: replace Abstract: Underwater environments pose unique challenges for robotic navigation and manipulation. While existing research has primarily focused on task-specif
arXiv:2506.00560v2 Announce Type: replace-cross Abstract: End-to-end planning systems for autonomous driving are rapidly improving, especially in closed-loop simulation environments like CARLA. Many s
arXiv:2601.00969v2 Announce Type: replace-cross Abstract: Vision-language-action (VLA) models provide strong action priors for robotic manipulation, but their reactive behavior can fail under distribu
v0.30.0-rc25 is a pre-release version of Ollama that changes the architecture to directly support llama.cpp instead of building on top of GGML, and allows for compatibility with GGUF file format. MLX
arXiv:2605.23504v1 Announce Type: cross Abstract: Anomaly detection in multivariate time series is a critical task across a wide range of real-world applications, where abnormal behaviour is rare, lab
arXiv:2605.23632v1 Announce Type: new Abstract: We introduce CopFITi, a copula model for probabilistic forecasting of irregular multivariate time series (IMTS). Our model combines the expressivity of
arXiv:2605.22851v1 Announce Type: cross Abstract: Photoplethysmography (PPG) has become a ubiquitous physiological signal; however, current generative models still struggle to preserve realistic wavef
arXiv:2605.23381v1 Announce Type: new Abstract: Though rectified flow models have achieved remarkable performance in image, video, and 3D generation, their practical deployments are challenged by slow
arXiv:2507.05064v4 Announce Type: replace-cross Abstract: Gaussian processes are flexible, probabilistic, non-parametric models widely used in machine learning and statistics. However, their scalabili
arXiv:2407.04573v4 Announce Type: replace-cross Abstract: Dense vector retrieval is an important building block of modern machine learning systems, underlying applications ranging from semantic search
arXiv:2605.22991v1 Announce Type: new Abstract: Reactive task-space planners such as Bug2 operate with fixed Cartesian step sizes and are unaware of the manipulator's joint-angle limits. When the Jaco
arXiv:2602.07399v2 Announce Type: replace Abstract: Vision--Language--Action (VLA) models bridge multimodal reasoning with physical control, but adapting them to new tasks with scarce demonstrations r
arXiv:2602.12579v2 Announce Type: replace-cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a dominant paradigm for enhancing Large Language Models (LLMs) reasoning,
arXiv:2605.22907v1 Announce Type: new Abstract: Real-world long video understanding requires models to perform continuous tracking, information integration and memory retention over massive temporal s
arXiv:2602.07801v4 Announce Type: replace-cross Abstract: In long-video understanding, conventional uniform frame sampling often fails to capture key visual evidence, leading to degraded performance a
arXiv:2605.23518v1 Announce Type: new Abstract: Directly editing ultra-high-resolution (UHR) images is valuable but underexplored, primarily due to the lack of high-quality data and the challenge in m
arXiv:2605.23141v1 Announce Type: new Abstract: A useful test of visual concept learning is not just whether a model can recognize a concept in a single image, but whether it can preserve and manipula