GPU 基础设施

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Learn how to deploy decision models with llama.cpp on Amazon SageMaker AI. This technical how-to serves OpenAI-compatible chat completions with the AWS Deep Learning Container for llama.cpp, then extends the runtime with SemIf to return probabilities over declared options on CPU or GPU.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Learn how to deploy one large language model across two GPU nodes on Amazon EKS using the multi-node Ray LLM Deep Learning Containers, with separate prefill and decode nodes behind one OpenAI-compatible endpoint. The pre-built container ships the full inference stack, so you configure the deployment instead of assembling it.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：As AI factories grow from megawatts to gigawatts, building them takes more than advanced computing. Open standards are helping the industry scale AI infrastructure faster, more efficiently and more reliably. At the Open Compute Project (OCP) Global Summit, taking place in the San Jose McEnery Convention Center, NVIDIA announced new partnerships and contributions aimed at […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Microsoft’s Maia 300 chip. | Image: Microsoft Microsoft is announcing its third-generation Maia in-house AI chip today, designed to be the "primary AI inference system" for Microsoft's products and services. Maia 300 not only improves the cost and efficiency of running AI workloads in the cloud, it's also more capable of handling the latest reasoning models and agentic tasks. Maia 300 will start being rolled out to production in 2027, and it has been designed with efficiency in mind. "We expect Maia 300 to deliver 40 percent more tokens per megawatt compared to leading GPU-based systems," says Scott Guthrie, executive vice president of Microsoft's cloud group. "And in a full 96-acceler … Read the full story at The Verge.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Advanced Micro Devices (AMD) designs processors and other semiconductor products used in data centers, personal computers, gaming systems and embedded devices. Its products also support the computing demands of AI applications. Its 2025 annual filing reported $34.6 billion in revenue, including $16.6 billion from its Data Center segment, up 32% from 2024. AMD’s investment in […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：TinyFish builds infrastructure for enterprises that need agents to work across the live web. Its Search, Fetch, Browser...

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.10845v1 Announce Type: new Abstract: A large language model can only use the text that fits in its context window, and it recomputes its internal key-value (KV) state for a prompt every time the prompt is sent. We test a memory layer, the public package galahad-kv, that saves the KV state of each block of about 16,000 tokens to encrypted local NVMe disk and loads it back later, byte-exact, without recomputing it. We ran it on 50,000,000 tokens of real public text, served through vLLM on one NVIDIA H100, with Gemma 4 12B and Gemma 4 31B. Every block we probed was loaded back from the encrypted store with no recompute (100 of 100, at depths from 0 to 50M tokens) on both models. Loading a block was 2.8x to 4.3x faster than recomputing it and used 8.8x t…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Turning a simulation idea into a working application means assembling assets, connecting physics and rendering, and checking that the scene behaves as intended. Developers are combining frontier AI models with NVIDIA Omniverse libraries to help carry out that work — building applications for exploring scenarios, investigating failures and improving designs. Developers direct AI agents through […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Turning a simulation idea into a working application means assembling assets, connecting physics and rendering, and checking that the scene behaves as intended. Developers are combining frontier AI models with NVIDIA Omniverse libraries to help carry out that work — building applications for exploring scenarios, investigating failures and improving designs. Developers direct AI agents through […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：A reference architecture for securely sharing one Amazon SageMaker HyperPod EKS cluster across multiple teams, using AWS IAM Identity Center for authentication, per-team SageMaker Domains and Kubernetes namespaces for isolation, HyperPod Task Governance for fairness, and namespace-level cost allocation for chargeback.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：The frenetic pace of AI development is challenging data center design and operational paradigms. The twin pillars of hyperscale computing The post AI data lakes are driving new storage demands appeared first on The New Stack.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Gears of War: E-Day leads the charge on GeForce NOW this week, bringing Marcus Fenix and Dom Santiago’s first fight against the Locust Horde to the cloud with GeForce RTX-powered performance. A new way to join the action is also coming: Fire TV users will soon be able to purchase GeForce NOW memberships directly through […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：At a Microsoft event in San Francisco on Wednesday, Jensen Huang and Satya Nadella outlined how NVIDIA and Microsoft are co-engineering hardware and software for AI agents to run on Windows PCs. NVIDIA was founded because of Windows, Huang said. Now AI agents are coming to Windows. “If you look at the entire journey of […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Microsoft just wrapped up a big Windows and Surface-focused keynote in San Francisco. The biggest announcement was arguably the release details about the Surface Laptop Ultra, its new laptop that’s powered by Nvidia’s RTX Spark Arm-based chip. The machine will start at $2,599 for a configuration with an 8-core CPU, 24GB of RAM, and 512GB of storage, and it will be released on October 16th. The laptop also has built-in magnetic USB-C charging instead of the proprietary Surface Connect magnetic charging port. The Verge‘s Tom Warren published a deep dive all about the new laptop. Microsoft also announced that the Surface RTX Spark Dev Box, a mini PC for developers, is available for preorder for $5,999 ahead of shipping in November. In addition, Microsoft shared de…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Microsoft's Nvidia-powered Surface RTX Spark Dev Box is available for preorder now directly, and slated to ship in November for just about $6,000. It's pricier than the DGX Spark mini PC Nvidia launched last year, but PC prices have been climbing due to shortages of RAM and other components. The Dev Box's flat, 3D-printed anodized aluminum chassis doubles as a heatsink and resembles the top vents on an Xbox Series X. It's launching alongside the new Surface Laptop Ultra and runs on Nvidia's Arm-based RTX Spark platform and 128GB of unified memory. With that much memory, along with a 100-watt thermal envelope and Nvidia's Tensor cores, you … Read the full story at The Verge.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.06963v1 Announce Type: new Abstract: Rotary Position Embedding (RoPE) encodes token positions by rotating each two-dimensional channel of the query and key vectors at a channel-specific frequency, making the attention logits invariant to a common shift of positions. However, this rotation is periodic, and it leads to position aliasing where relative positions separated by a full rotation period become hard to tell apart. To address this, we propose WavePrune, which restricts each channel to its first rotation period. We show that it removes the distractions in attention maps created by position aliasing and improves overall long-context performance. Specifically, WavePrune raises the HELMET score on four of five models we test without any extra tunin…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.06890v1 Announce Type: new Abstract: We present an industry experience report on three years of operating an event-driven cloud infrastructure for continuous machine learning training in automotive manufacturing. Our system orchestrates GPU-accelerated training of product-specialized model pairs, a physics prediction model and a reinforcement-learning control policy, across multiple plants, coordinating long-running GPU workloads triggered by manufacturing events. The architecture combines Amazon ECS with EC2 GPU capacity providers, SQS-based messaging with dead-letter queues, and an admission-controlled Lambda dispatcher that enforces cluster concurrency limits. A Conductor orchestrator on ECS Fargate initiates dependency-aware retraining chains on…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.06861v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has become the dominant paradigm for eliciting multi-step reasoning in large language models, and a recent wave of methods (LUFFY, ExPO, PAPO, TAPO) further augments RL with \emph{external guidance} - expert traces, self-explanations, or retrieved thought patterns. Although each method reports empirical gains, none provides convergence rates, bias bounds, or an optimal weighting rule for the guidance signal. We close this gap with \emph{Guidance-Augmented GRPO} (GA-GRPO), a unified theoretical framework that casts external guidance as a stochastic guidance operator G re-writing the question distribution, and analyses the resulting policy-gradient estimator as a…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.06917v1 Announce Type: new Abstract: Prefill-decode disaggregation is becoming a common architecture for LLM serving because it separates two phases with distinct execution patterns and SLO objectives. Existing systems typically combine a fixed prefill/decode worker ratio with request routing across workers. However, real-world workloads exhibit both short bursts and sustained shifts in the prefill-to-decode demand ratio. As a result, a configuration that is well provisioned at one time may quickly become mismatched, causing latency SLO violations even when idle capacity exists elsewhere. Existing autoscaling mechanisms can add capacity, but they react slowly, require spare GPUs, and do not directly address short-timescale phase imbalance. We present…

Mistral 放出 Mistral Large 4 预览版：总参数 1 万亿、激活参数 490 亿，在自建的 3,800 块 NVIDIA Grace Blackwell GPU 集群上训练。API 预览版已上线，开放权重承诺本月底发布；Artificial Analysis 得分 38，较上一代 Large 3 的 9 分大幅跃升，但整体仍落后前沿约 6 个月。

Mistral AI 以公开预览形式发布 Mistral Large 4（内部代号 Le Chonk）：细粒度 MoE，总参数 1.05 万亿、每 token 激活 490 亿，配 16 亿参数视觉编码器与 100 万 token 上下文，在欧洲自有数据中心用 3800 块 NVIDIA Grace Blackwell GPU 从零训练。API 已上线，输入/输出每百万 token 1.36/4.18 美元，缓存输入 0.14 美元；权重与许可证预计 2026 年 10 月底公布，暂不能自托管。最突出成绩在网络安全：Cybench 93%、CyberGym-E2E 82%，Mistral 称多家闭源前沿模型因拒绝任务而接近零分。

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Campaign groups report surge in membership after a spate of AI safety alerts and apocalyptic warnings In a bar below Waterloo Bridge, a cell of a fast-growing anti-AI protest movement met last week to plot their latest move: disrupting a tech industry dinner being addressed by a senior executive from the chip maker Nvidia. Emma, a 28-year-old bartender, had never before taken direct action and was the most nervous of the four plotters from Pull The Plug, one of a growing number of AI-focused campaign groups around the world. Continue reading...

Together AI 正与 IBM 和 NVIDIA 合作，在 IBM Cloud 上部署首个专用大规模 NVIDIA B300 GPU 推理集群，并采用 Spectrum-X 以太网网络，以扩展企业级开放模型推理。

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Breast cancer is the most commonly diagnosed cancer among American women — yet the gaps in care are wide. A majority of women over age 40 skip the recommended annual screening. Radiologists are reading more mammograms with fewer colleagues. And when a diagnosis arrives, the tests that inform treatment can take weeks to return results. […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.02248v1 Announce Type: new Abstract: Operational land surface forecasting systems built on Mamba-family Structured State Space Models absorb non-stationary confounding events (unrecorded irrigation booms, dam-operation shifts, sensor recalibrations) into their state-transition matrices, silently biasing NDVI, LST, and crop phenology predictions long after the physical cause ends. This paper introduces SSU-LSF (State-Space Unlearning for Land Surface Forecasting), the first machine-unlearning framework purpose-built for geoscientific Mamba-based SSMs. We develop EKFac influence functions specialized to the Mamba state matrices via a closed-form matrix-exponential gradient, use spectral-radius-weighted elbow thresholding to localize a temporal confound…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Cantina Security, with Yeta Labs, has released apex-flash-1, an open-weights model trained specifically for vulnerability research. It is a reinforcement learning fine-tune of Z.ai’s GLM-5.3-Flash, released on Hugging Face under the MIT license. Is it deployable? Yes, the MIT weights serve on vLLM, SGLang or Transformers, but BF16 needs roughly 640 GB of GPU memory. […] The post Can an Open Model Do Security Research? Cantina’s apex-flash-1 Solves 40 of 60 Held-Out Bug Tasks appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Inside OpenAI’s DevDay releases, Gemini 4 Argon, Meta’s enterprise ambitions, AMD’s World Labs deal, and Instinct’s billion-dollar round.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Aleph Alpha has released Kolibri, a 78.1B-parameter English-German Mixture-of-Experts model that activates only 3.46B parameters per token. It has a 1M-token context and per-request reasoning effort, and its Apache 2.0 FP8 weights run on a single B200 or H200. The post Aleph Alpha Releases Kolibri: A 78.1B Open-Weight English-German MoE Model With Only 3.46B Active Parameters appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Prime Intellect has launched Prime Inference, an OpenAI-compatible platform for serving frontier open models on NVIDIA Blackwell. Its GLM-5.3 deployment uses Dynamo, vLLM and NVFP4 KV compression to serve 66 sessions per prefill group at 101 tok/s per user. The post Prime Intellect Launches Prime Inference: Serverless and Reserved Serving for Frontier Open Models appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.00053v1 Announce Type: new Abstract: Four-bit floating-point (FP4) Tensor Cores accelerate matrix multiplication, but scale computation, operand packing, layout construction, and saved backward state can erase the gain. We present \emph{format-aware fusion}, which co-designs each quantization producer with its scale domain and consumer layout for native \mxfp{}, global \nvfp{}, and cooperative-thread-array-local \nvfp{}. We evaluate Llama-3-family 8B pretraining through 160 billion tokens using bfloat16 output projections and compiled cross entropy. In matched same-accelerator probes, bfloat16 and Transformer Engine \nvfp{} reach 18.8K and 27.6K tokens/s/GPU, while our fastest custom route reaches 37.9K. \mxfp{} with row-gradient stochastic rounding…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.00049v1 Announce Type: new Abstract: Graphics processing unit (GPU) generations scale matrix, special-function, and memory pipelines at different rates, so kernel bottlenecks move as hardware evolves. FlashAttention-4 exposed this imbalance inside attention on NVIDIA Blackwell. We test whether short polynomial programs can accelerate other special-function-unit (SFU) operations in large language models (LLMs). We first compare native PyTorch evaluation with packed fused multiply--add (FMA) programs in an isolated IEEE binary16 (FP16) sweep spanning L2-resident and high-bandwidth-memory (HBM)-resident working sets. We then replace native sigmoid, tanh, and sigmoid linear unit (SiLU) with degree-3 or degree-4 bfloat16 (BF16) programs in four GB200 inte…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：NVIDIA announced a new 64GB configuration of DGX Spark — from Acer, ASUS, Dell, Gigabyte, HP and MSI — its GB10-powered desktop AI system. It gives developers a way to start with one system for local models and agents, then cluster two 64GB units for 128GB of memory across the cluster and more compute when […] The post NVIDIA Announces DGX Spark 64GB: A 1-PetaFLOP Grace Blackwell Desktop for Local AI Agents, Fine-Tuning, and Inference appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Local AI is becoming more useful by the token. As AI agents move from experiments into everyday development, increasingly capable open models are shrinking to fit on more devices, giving builders more to run locally. Coming this month, NVIDIA DGX Spark will be available with 64GB of unified memory from top manufacturer partners — Acer, […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2610.00355v1 Announce Type: new Abstract: Efficient indoor LiDAR perception is challenging because mobile robots must understand cluttered three-dimensional environments under strict latency and memory constraints. Existing point-based and voxel-based methods often incur substantial computational overhead, whereas conventional bird's-eye-view (BEV) representations improve efficiency at the cost of discarding vertical geometric information. We present IndoorBEV, a lightweight LiDAR perception framework that mitigates this tradeoff through a height-aware BEV representation and geometry-conditioned feature fusion. IndoorBEV summarizes the vertical point distribution in each BEV cell using statistical height features and multi-frequency height encoding, allow…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2609.38198v1 Announce Type: new Abstract: Diffusion maps, and kernel methods more generally, provide an interpretable nonlinear spectral representation basis for geometric learning. In the geometric limit, small bandwidth, these matrices tend to be high rank and thus require materializing dense Gaussian kernels requires $O(N^2)$ memory. We introduce FlashDiffusion, a matrix-free method that evaluates dense Gaussian kernel blocks in fused GPU tiles and couples the eigensolver to an empirical $\beta$-flow that selects the finite-sample resolution scale. A continuation over sample size and bandwidth warm-starts increasingly expensive spectral solves from coarser resolutions.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：The AI chipmaker’s new safety platform adds controls around AI agents, while enterprises remain responsible for defining their authority and setting boundaries.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：GPT-6 Astra Ultrafast, running on NVIDIA Blackwell GPUs, is available now in the OpenAI API and to eligible ChatGPT Work and Codex users. Accelerated by inference optimizations through OpenAI’s models that tap into the capabilities of the NVIDIA Blackwell architecture, Ultrafast offers up to 8x faster token generation than the Astra Standard mode. For developers, […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Learn how to use Amazon S3 Vectors as the persistent memory layer within the NVIDIA NeMo Agent Toolkit (NAT), deployed on Amazon Elastic Kubernetes Service (Amazon EKS). This post shows how NAT's memory subsystem works and how to implement Amazon S3 Vectors as a custom memory provider, using a multi-agent investment research use case.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Today I’m talking with Josh Dzieza, a longtime features writer here at The Verge, about Kevin O’Leary’s plans to build a massive data center in Utah. The idea was to build the world’s biggest data center — a 40,000-acre AI campus with nine gigawatts of power, or more than double the average power usage of the entire state of Utah. The project is technically called Stratos, but it’s more prominently known as Wonder Valley, a reference to O’Leary’s nickname on Shark Tank. Josh has spent months reporting on this project, and it’s fair to say Wonder Valley has completely upended Utah politics. What Josh found throughout the course of his reporting was that the way this data center came together — how it was planned, how it was announced and approved, and how local…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：AI factories are built by the megawatt, even by the gigawatt. Each megawatt factory costs roughly $60 million, and AI factory operators will only commit capital on that scale with a clear view of the return on investment. Three key things shape AI factory returns: Earning capacity: What the factory could earn in a year […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Multi-node GPU clusters with RDMA, gang scheduled from Modal's shared capacity pool and billed by the second, behind a single decorator.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：NVIDIA has released Kumo Tabular, a new family of tabular foundation models (TFMs) for classification and regression. If you have followed TabPFN or TabICL, the setup will look familiar. The model takes labeled rows as context and predicts new rows in one forward pass. There is no training, no hyperparameter tuning, and no feature engineering. […] The post NVIDIA Releases Kumo Tabular: Open Tabular Foundation Models That Predict New Rows in a Single Forward Pass appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2609.38216v1 Announce Type: new Abstract: Existing benchmarks evaluate tabletop manipulation, flat-floor household activity, or humanoid locomotion and manipulation as separate task groups; none scores vertical mobility and dexterous work on a fragile payload in one long-horizon episode. We present Fiatlux, a light-bulb replacement benchmark built on NVIDIA Isaac Lab. In one episode, a Unitree G1 humanoid positions a step ladder under a ceiling or wall fixture, climbs it, exchanges a spent bulb in a socket for a fresh one, and leaves the spent one in a disposal crate. We decompose the episode into twelve subtask environments scored on difficulty-weighted gates. The goal is a successful replacement, with the fresh bulb seated, the spent one disposed of, ne…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Bringing together the world’s brightest minds and the latest accelerated computing technology leads to powerful breakthroughs that help tackle some of the biggest research problems. To foster such innovation, the NVIDIA Graduate Fellowship Program provides grants, mentors and technical support to doctoral students doing outstanding research relevant to NVIDIA technologies. The program, in its 26th […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Amazon Bedrock AgentCore Runtime Instances gives multi-agent workflows AWS managed EC2 infrastructure with GPUs, persistent volumes, and multi-day sessions. In this post, we deploy a three-agent music production pipeline where the agents colocate on one GPU instance, share a filesystem, and hand work to each other to produce a finished track.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Building on nearly a decade of co-engineering, CoreWeave has built NVIDIA compute, networking and software into a cloud purpose-built for AI that’s still returning on investment across multiple generations of deployment. Now, CoreWeave is bringing the next generation of NVIDIA infrastructure to production. At CoreWeave Fully Connected, running this week in San Francisco, CoreWeave announced […]

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Anyone else getting “The Last Supper” vibes from this photo taken at the White House tech gathering? | Photographer: Tierney L. Cross/The Washington Post/Bloomberg via Getty Images We now have the full details of the "morally binding" AI safety deal announced by President Trump yesterday, in which top executives agreed to self-regulate their artificial intelligence technology. The accord, officially titled the Joint Commitment On Frontier Responsibilities, was shared online by tech founder and presidential advisor David Sacks, and has been signed by Google's Sundar Pichai, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, XAI's Elon Musk, and Nvidia's Jensen Huang. "In order to build a positive future for the American people and the worl…

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：OpenClaw has come a long way since it emerged as a viral weekend project less than a year ago. Created The post “Think of it as Kubernetes for agents”: OpenClaw lands in the enterprise with Nvidia and Red Hat on board appeared first on The New Stack.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：Video world models can render convincing clips that still break physics. Butter spreads like paint. Balls pass through walls. A team from NVIDIA, MIT and the University of Oxford argues the fix can come from language itself, not from extra visual, latent or numerical signals. Their framework, Physis-Lang, treats physical language as a shared, optimizable […] The post NVIDIA Researchers Introduce Physis-Lang: Self-Evolving Physical Language That Lifts Cosmos 3 Past Veo 3.1 on Physics Benchmarks appeared first on MarkTechPost.

AI 服务暂时不可用，以下为来源摘要，待恢复后补全翻译：arXiv:2609.36134v1 Announce Type: new Abstract: Functional Kolmogorov-Arnold Networks (FunKAN) achieve state-of-the-art accuracy on MRI Gibbs artifact removal and anatomical segmentation, but their 11.6 M parameters and 8.7 GFLOPs are too large for edge medical devices. We present FunKANLite, a two-stage, hardware-aware compression of FunKAN for point-of-care use. FunKANLite-TR reduces the spatial prior and replaces the ResBlock offset predictor with a depthwise-separable block. It has 1.9x fewer parameters than FunKAN and no loss in accuracy. We then distill FunKANLite-TR into FunKANLite-ST, which lowers the Hermite basis rank, factorizes the spatial prior into a low-rank form, and halves the filter widths. FunKANLite-ST has 5.6x fewer parameters and 3.7x fewe…