local ai

arXiv:2608.06819v1 Announce Type: cross Abstract: Token-level collaboration allows a large language model (LLM) to assist a small language model (SLM) when their predictions diverge. Existing methods

arXiv:2608.07116v1 Announce Type: cross Abstract: Camera localization in bronchoscopy remains a challenging problem due to stringent accuracy requirements, real-time constraints, and limited training

arXiv:2608.06886v1 Announce Type: new Abstract: Dehazing networks are commonly trained on synthetic hazy-clear pairs, but their performance often drops on real photographs. Synthetic haze generated us

arXiv:2608.02254v2 Announce Type: replace Abstract: exttt{Homebot} is a locally deployable AI agent for conversational household assistance and automation. It accepts voice and instant-messaging reque

What’s the best way to recommend products to little-known users? We’ve spent our careers trying to solve this problem for major companies like Spotify and Priceline, and it’s why Sidd founded Malachyt

I'm playing with a conversational agent I made using either api/generate or api/chats. In both case I do ask him to not ask follow up question, to not act like an assistant, etc. Either from a system

arXiv:2608.07141v1 Announce Type: new Abstract: Evaluating the perceptual alignment between Contrastive Vision-Language Models (CVLMs) and humans is typically constrained by traditional benchmarks tha

arXiv:2608.07091v1 Announce Type: cross Abstract: Edge Artificial Intelligence (Edge AI) enables the deployment of AI models directly on local edge devices, while such deployments are subject to stric

Update : New version is ready and online , should be much better, fully functionnal on ComfyUI, and you can find before/after here : https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA/blob/main

arXiv:2608.06402v1 Announce Type: new Abstract: Community detection is a fundamental task in graph analytics that aims to identify cohesive groups of entities with similar behaviors or interests. Clas

arXiv:2608.07074v1 Announce Type: new Abstract: Dense point cloud maps, as a typically used mapping representation, are difficult to deploy on resource-constrained robots because their memory consumpt

arXiv:2608.07192v1 Announce Type: new Abstract: Low-resolution infrared (IR) array sensors represent an interesting solution for privacy-preserving human sensing in embedded systems. In this letter, w

arXiv:2608.06916v1 Announce Type: new Abstract: Quantized Neural Networks~(QNN) with low-bitwidth data have proven promising in efficient storage and computation on edge devices. To mitigate accuracy

MiniMax H3 loads a 32B text encoder, 15.7 GB, just to turn your prompt into a conditioning tensor. I replaced it with a Qwen3-VL 4B or 8B plus a learned map into the same space. Same DiT, same VAEs, s

Muse Glimmer, the new 30B model, is available on Hugging Face right now - here's the GGUF version: https://huggingface.co/meta-models/Muse-Glimmer-30B-GGUF 1/ big announcement today: we will be releas

VLX-Seek-1.5-10B VLX-Seek-1.5-10B is the open-source 10B model in the VLX-Seek 1.5 family, designed for fine-grained perception and visual grounding in embodied scenarios. It targets practical setting

arXiv:2608.06701v1 Announce Type: cross Abstract: Fixing GitHub issues in large-scale projects is a long-horizon task, especially when a fix requires changes across multiple locations or the issue des

arXiv:2507.07814v2 Announce Type: replace Abstract: We introduce a novel upper bound on the local Lipschitz constant of the dot-product self-attention block showing its dependence on the attention map

I built myself a personal AI art history assistant https://github.com/lololerigolo60/RAG-art/tree/main I love art history but I have way too many books, PDFs, and notes scattered everywhere. So I buil

Söker en riktigt vass programmerare – jag har ett projekt jag tror kan bli stort. Jag letar efter en extremt kunnig utvecklare som vill hoppa på ett projekt från ett tidigt skede. Jag kan inte avslöja

arXiv:2608.06946v1 Announce Type: cross Abstract: In gossip learning, a network of nodes trains a shared model collaboratively, without a central coordinator, by repeatedly exchanging parts of their l

v0.32.8 is an Oct 10, 2023 release of the ollama repository on GitHub, following a pre‑release tag v0.32.8‑rc0. The update adds Muse Glimmer support for NVIDIA, AMD and additional platforms, with the

As promised, the first Batch of Characters that Minimax knows - American knowdledge Edition. Hope this helps the Community. Workflow for this was simple: This is the Prompt: Brad Pitt integrated_multi

Here is the second Edition, Videogames. Workflow is the same as in the first Part, its pretty simple: This is the Prompt: Brad Pitt integrated_multimodal_description: [Shot 1] Live-action, contemporar

Spec-dec has been a thing for a while, in fact, it's wasn't an idea that was born for LLM inference. E.g. Uber's https://github.com/uber/submitqueue applied it to a merge queue. Apple & GDM had been r

Francesco @francedot reported that the first fully offline computer‑based LLM experience was achieved by running Muse Glimmer 30B (from @AIatMeta) locally on macOS, using Cua Driver to control native

arXiv:2608.07033v1 Announce Type: new Abstract: This work investigates whether Electroencephalograph (EEG) foundation models (EFMs) can be made faster and locally deployable without sacrificing accura

Ollama allons you to import model but have you ever tried doing so ? Like running model imported from hugging face or you own model ? Any use case you wanna share ? Very curious about that submitted b

Former Ollama Cloud $20 dollar plan holder look at returning. How's the state of the usage ATM? It was in a dire state when I left a few months ago. Is it still very limited with Mid sized models? M3,

TL;DR: I am building a Windows-first local AI setup for people who want to try local LLMs without spending days choosing models, setting up Ollama, Docker, WSL, Open WebUI, agents, and tool permission

Who remembers when OpenAI previewed Sora two years ago and the quality felt unreal? We had never seen anything like it. Back then, Sora 1 didn't even generate audio and was heavily censored. Prompt: i

Hi folks, I hate slop as much as you do, so instead of starting with 'The Problem', I'll just cut to the chase: I just published Lophius, which is the culmination of more than two years of fighting wi

H3 weights went live on HuggingFace August 3rd and I started pulling them immediately. An omni-modal video model with native stereo audio in the same forward pass, where audio can actually drive the v

Would it be possible and make sense to add metadata to training data e.g. a trustfactor (0.0 - 1.0)? For example: the older data is the less trustworthy it is. And data after 2022 gets less trustworth

Shane Burke / The Information: US data center bans top 500, up from 300+ in late June, as New York and Texas join cities and counties pushing back against data center development — Local government re

Over the past few months I have been building a CPU-first inference engine from scratch in pure C99 (no Python, no CUDA, no BLAS, just GCC and make). The focus has been running 1.58-bit ternary models

I discovered recently that Google uses their own TPUs, like tiny ASIC cards like the toy ones that existed for bitcoin. And while it sounds inefficient the fact they use thousands of them because...th

What IDE are you using. its another problem area for me . I usually use VSCode , but with local llms I have not found an extension which works optimally VSCode CoPilot chat with Ollama: CoPilot bloats

seems to be about as good as a vega 56 with 16Gb of VRAM, is it worth it? (don’t want to deal with NVIDIA drivers on Linux, already have an rx6650xt and might simply use vulkan for llamacpp inference)

Hi everyone, I'm a systems analysis student researching a problem a lot of you probably know well: how much you actually trust the published VRAM/RAM requirements for open-source models before trying

MiniMax-H3 in ComfyUI 0.30.0, RTX 4080 16 GB (224 W cap). int8 DiT + int8 Qwen3-VL-32B text encoder. Turbo LoRA @ 0.9, euler + simple, 6 steps, no CFG. MiniMaxH3ReferenceToVideo with 3 reference image

I have not been successful with management to get funding for local resources despite bringing forth solid arguments about data sovereignty and related architectures. What actually succeeded in gettin

arXiv:2608.05784v1 Announce Type: new Abstract: Computer-use agents pay full frontier inference to re-derive routines their user has already performed, because an agent's memory today records what the

Press My earlier prediction that Tesla would buy them completely missed the mark. With AMD focusing heavily on the enterprise side, the idea of consumer-facing hot-swappable AI model chips looks prett

arXiv:2608.05232v1 Announce Type: new Abstract: The work of Tang et. al. (2025) on numerical translation is extended by analysing the capability of five large language models (LLMs) for the localisati

arXiv:2608.05880v1 Announce Type: cross Abstract: Interpreting clustering outcomes remains a fundamental challenge in data analysis, particularly in domains such as healthcare where meaningful pattern