Knowledge catalogue
We built our internal AI engineering stack on the same products we ship. That means 20 million requests routed through AI Gateway, 241 billion tokens processed, and inference running on Workers AI, se
arXiv:2604.14853v1 Announce Type: new Abstract: Test-time compute scaling, the practice of spending extra computation during inference via repeated sampling, search, or extended reasoning, has become
arXiv:2604.14528v1 Announce Type: cross Abstract: Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains p
arXiv:2604.14388v1 Announce Type: new Abstract: Humans routinely infer taste, smell, texture, and even sound from food images a phenomenon well studied in cognitive science. However, prior vision lang
Amazon Bedrock now supports cost allocation by IAM principal (users and roles) in AWS Cost and Usage Report 2.0 and Cost Explorer, enabling customers to understand and attribute inference costs across
Running LLMs across Cloudflare’s network requires us to be smarter and more efficient about GPU memory bandwidth. That’s why we developed Unweight, a lossless inference-time compression system that ac
arXiv:2604.13226v1 Announce Type: new Abstract: Large Language Models (LLMs) rely heavily on Key-Value (KV) caching to minimize inference latency. However, standard KV caches are context-dependent: re
arXiv:2604.13291v1 Announce Type: new Abstract: Accurate characterization of subsurface heterogeneity is challenging but essential for applications such as reservoir pressure management, geothermal en
انتهت حجة 'ما عندي كرت شاشة قوي عشان أشغل ذكاء اصطناعي محلي'. علي بابا نزلت موديل Qwen3.6 بهندسة MoE. حجمه الكلي 35B (يعطيك ذكاء عالي جداً)، لكن وقت الـ Inference يستهلك 3B فقط! يعني أداء Enterprise ع
arXiv:2604.12719v1 Announce Type: new Abstract: The deployment of deep neural networks in safety-critical systems necessitates reliable and efficient uncertainty quantification (UQ). A practical and w
arXiv:2604.12852v1 Announce Type: new Abstract: Collaborative transport requires robots to infer partner intent through physical interaction while maintaining stable loco-manipulation. This becomes pa
arXiv:2604.12247v1 Announce Type: cross Abstract: Speculative decoding has emerged as a promising approach to accelerate autoregressive inference in large language models (LLMs). Self-draft methods, w
arXiv:2604.09686v1 Announce Type: new Abstract: Traditional neural network models for intent inference rely heavily on observable states and struggle to generalize across diverse tasks and dynamic env
arXiv:2604.09612v1 Announce Type: new Abstract: Theory of Mind, the capacity to explain and predict behavior by inferring hidden mental states, has become the dominant paradigm for social interaction
arXiv:2604.09611v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used in applications forming multi-request workflows like document summarization, search-based copilots,
arXiv:2604.10815v1 Announce Type: cross Abstract: MeloTune is an iPhone-deployed music agent that instantiates the Mesh Memory Protocol (MMP) and Symbolic-Vector Attention Fusion (SVAF) as a productio
arXiv:2604.10401v1 Announce Type: new Abstract: Inferring nationality from personal names is a critical capability for equity and bias monitoring, personalization, and a valuable tool in biomedical an
arXiv:2604.09557v1 Announce Type: cross Abstract: Speculative Decoding (SD) has emerged as a critical technique for accelerating Large Language Model (LLM) inference. Unlike deterministic system optim
arXiv:2509.20868v2 Announce Type: replace-cross Abstract: Structured reasoning can improve the inference performance of large language models (LLMs), but it also introduces computational cost and cont
**b8771** is a build release of [llama.cpp](https://github.com/ggml-org/llama.cpp), the open-source C/C++ library for running large language model (LLM) inference locally. Like other incremental build
arXiv:2506.17788v2 Announce Type: replace Abstract: Social reasoning - inferring unobservable beliefs and intentions from partial observations of other agents - remains a challenging task for large la
arXiv:2604.08592v1 Announce Type: new Abstract: Reservoir observers provide a data-driven approach to the inference of unmeasured variables from observed ones for nonlinear dynamical systems. While pr
MiniMax-M2 is a large-scale mixture-of-experts (MoE) language model available for inference on Together AI's platform, accessible via both serverless and dedicated infrastructure options. The model ca
arXiv:2604.07622v1 Announce Type: new Abstract: Speculative decoding is an effective technique for accelerating large language model inference by drafting multiple tokens in parallel. In practice, its
arXiv:2604.08075v1 Announce Type: new Abstract: Production vLLM fleets typically provision each instance for the worst-case context length, leading to substantial KV-cache over-allocation and under-ut
arXiv:2604.06262v1 Announce Type: cross Abstract: Contextual clinical reasoning demands robust inference grounded in complex, heterogeneous clinical records. While state-of-the-art fine-tuning, in-con
arXiv:2604.07812v1 Announce Type: new Abstract: In multimodal large language models (MLLMs), the surge of visual tokens significantly increases the inference time and computational overhead, making th
arXiv:2604.06260v1 Announce Type: cross Abstract: Test-time scaling investigates whether a fixed diffusion language model (DLM) can generate better outputs when given more inference compute, without a
Microsoft Foundry Local is now generally available as an end-to-end local AI solution that enables developers to bring AI inference directly into their applications with no cloud dependency, no net...
NVIDIA, Google DeepMind, EMBL-EBI, and Seoul National University collaborated to extend the AlphaFold Protein Structure Database (AFDB) beyond monomeric structures to proteome-scale quaternary stru...
arXiv:2610.20832v1 Announce Type: cross Abstract: Despite their extensive pretraining knowledge and strong reasoning capabilities, large language models (LLMs) struggle to adaptively gather informatio
arXiv:2602.13832v2 Announce Type: replace Abstract: Large language model (LLM) agents are increasingly capable of carrying out complex, long-horizon tasks, yet effective assistance requires more than
arXiv:2610.21565v1 Announce Type: cross Abstract: On-device deployment of large language models (LLMs) is increasingly adopted due to latency and privacy requirements, but exposes proprietary models t
arXiv:2610.22143v1 Announce Type: new Abstract: A perception system that must brake for a pedestrian does not need a low average latency. It must deliver an answer before the deadline on every frame,
arXiv:2601.21132v2 Announce Type: replace Abstract: I demonstrate that large language models can infer ethnicity from names with accuracy exceeding that of Bayesian Improved Surname Geocoding (BISG) w
arXiv:2610.21977v1 Announce Type: new Abstract: Physics-informed Gaussian processes (GPs) provide a principled framework for scientific inference by expressing uncertainty over empirical measurements,
arXiv:2610.21220v1 Announce Type: cross Abstract: Recommender systems often need to balance multiple conflicting objectives. Controllable Multi-Objective Recommendation (CMOR) enables dynamic adjustme
arXiv:2610.21695v1 Announce Type: new Abstract: Fixed-function CNN accelerators make edge inference fast and energy-efficient, but they execute only networks that obey a precise numerical contract. On
arXiv:2610.20669v1 Announce Type: new Abstract: Reward-guided discrete diffusion enables test-time alignment with downstream objectives, but applying Sequential Monte Carlo (SMC) guidance at every den
arXiv:2610.15298v1 Announce Type: new Abstract: Large language models (LLMs) offer a new approach to wearable-sensor-based human activity recognition. Retrieval-augmented generation (RAG) further intr
arXiv:2610.19957v1 Announce Type: new Abstract: World action models (WAMs) jointly predict future visual states and robot actions, but processing dense future-frame tokens incurs substantial inference
arXiv:2610.17831v1 Announce Type: cross Abstract: Diffusion large language models (dLLMs) are appealing for code generation because they can update many tokens in parallel, yet many are optimized for
arXiv:2610.17816v1 Announce Type: new Abstract: Edge-cloud collaboration combines the lower inference overhead and greater data locality of edge-side execution with the stronger capabilities of cloud-
arXiv:2610.14970v1 Announce Type: new Abstract: With the rapid evolution of large language models (LLMs), there is an increasing demand for effective personalization techniques that can adapt model be
arXiv:2610.19114v1 Announce Type: new Abstract: Activation steering controls large language models at inference time, but its effectiveness depends on the choice of reference data. We introduce genesi
Editor's note: Generative AI startups are moving rapidly beyond static clip generation to build interactive, real-time worlds. In this technical deep dive, we explore how AI startup Visko collaborated
arXiv:2610.19765v1 Announce Type: new Abstract: Computationally lightweight manipulation policies predict executable actions with low inference latency, but prediction errors can limit task success. I
arXiv:2610.17305v1 Announce Type: new Abstract: Large Language Models (LLMs) have recently improved Text-to-SQL generation, but their computational cost can hinder deployment in resource-constrained s
arXiv:2610.19232v1 Announce Type: new Abstract: Large language models impose substantial memory demands and arithmetic costs during inference. Dense-to-mixture-of-experts (MoE) conversion is a model o
arXiv:2610.20390v1 Announce Type: new Abstract: With the rapid development of vision-language models (VLMs), Web-GUI agents have achieved promising performance in interacting with real-world interface
arXiv:2610.14252v1 Announce Type: new Abstract: Looped transformers reuse the same layer multiple times, allowing inference compute to be adjusted without changing the model. Recent adaptive-depth met
arXiv:2610.18018v1 Announce Type: new Abstract: Low-light image enhancement (LLIE) is an important image/video processing problem for reliable visual systems under poor illumination, yet it remains ch
arXiv:2610.16118v1 Announce Type: new Abstract: Hour-long videos, streaming feeds, and movie-length content are pushing vision-language models (VLMs) past their breaking point -- the standard recipe o
arXiv:2610.15933v1 Announce Type: new Abstract: Predictive coding networks (PCNs) promise properties not offered by backpropagation-trained networks: learning from purely local information, a single n
arXiv:2610.16289v1 Announce Type: new Abstract: Quantization has emerged as one of the primary techniques for reducing the memory footprint of large language models during inference. A prominent examp
Inspired by TypeSafe's Jev (a non-generative model that only outputs bounded decisions — choices, probabilities, calibrated confidence), I built two small decision models end-to-end on identical budge
arXiv:2610.18825v1 Announce Type: cross Abstract: Simulation-based Bayesian inference for Chemical Reaction Networks (CRNs) must simultaneously solve two computationally demanding problems: discrimina
arXiv:2610.14703v1 Announce Type: new Abstract: On-device inference imposes constraints on model storage and operand traffic and thus benefits from compressed operand representations, which mandate th
arXiv:2610.17141v1 Announce Type: new Abstract: World Action Models (WAMs) have emerged as a promising alternative for embodied intelligence by jointly predicting future world states and robot actions
arXiv:2610.15638v1 Announce Type: new Abstract: Knowledge graph (KG) foundation models (KGFMs) predict links on unseen graphs zero-shot. KGFMs have been pioneered by ULTRA which builds a relation grap