Knowledge catalogue
arXiv:2609.20989v1 Announce Type: cross Abstract: As generative AI is increasingly used as a source of personal financial guidance, understanding how people appraise such advice is important for suppo
I shipped something I've been building for the last few weeks : phantom-kv , a refusal-removal system for large language models that doesn't touch a single weight. Instead of editing the model, it loa
arXiv:2609.21225v1 Announce Type: new Abstract: Parametric CAD reconstruction requires recovering both precise geometry and editable modeling operations from visual observations, making it challenging
arXiv:2609.21804v1 Announce Type: new Abstract: Given a compact semantic scene graph, long-term indoor video relocalization estimates a map-frame trajectory after lighting and furniture changes. Visua
Xiaomi: Xiaomi debuts open-weight omnimodal models MiMo-V2.6 Pro and Flash; Pro allegedly performs “on par with Opus 5 and GPT-5.6 Sol across most agent benchmarks” — Frontier intelligence, all the mo
OpenAI’s decision to retire custom GPTs appears to rest on the assumption that Plugins, Skills, Projects, Apps, and Workspace Agents collectively replace their functionality. For many Small and Medium
Doc + Jev => DocJev @llama_index 团队最新开源文档分类与拆分工具 DocJev，给它一个 PDF/DOCX/PPTX 和一份用自然语言写的类别规则，它可以告诉你这份文档属于哪个类别，或者把一份“文件包”按边界拆成若干子文档并给出各自页码范围。 开源地址：https://github.com/jerryjliu/docjev 核心架构：解析与决策分离 1. 解析层（O
Hey everyone, After wrestling with hardware and engine issues for days, I finally got Qwen 3.8 Next running properly on my multi-GPU rig. Thought I’d share the setup journey, benchmarks, and thermal r
I gave Jev, Laya, a finetuned ModernCE-base-nli and a finetuned Qwen3.5-4B the controls to Doom. Thanks to TypeSafe AI for Jev access. The video lines up the starts of four separate games with the sam
My comment on MCP was always a bad idea? — Hacker News.This article entirely misses the value that MCP brings today. Sure, there's almost no reason to use MCPs if you are running a full-blown terminal
Hello! So, yesterday i tried out Gemma 4 26B A4B QAT Q4_K_XL from here with Q4_0 MTP model in Pi agent. Suddenly, it became the first model to actually pass my simple test with HTTP server in C++, whi
So my brother and I both use LLMs for coding. I've started using a local GLM 5.3 Flash instance - q4 qat. My brother uses GPT-6-Astra as his daily driver. He has mentioned repeatedly to me that his ap
TL;DR: Local agent loop, ~21 days, one RTX 3090. Task was pretty much 'build a CUDA inference engine for optimized for yourself on this GPU arch.' Got working kernels and benches, not a win over llama
Josipa Majic Predin / Forbes: Vercel, Cloudflare, and others quickly add Jev, as it makes AI tool selection much faster and cheaper; TypeSafe: Jev matches GPT-5.6 and Sonnet 5 workflow evals — Most of
Just find it interesting, since if Taalas tried to get into the consumer market, it could be interesting. Kind of like how we have game discs on CDs. Of course, a Qwen3.8-27B chip is stuck on 3.8-27B,
Screenshot of heavy load. I did offload vision to a spare 2060 instead of local ram. The control panel was generated with qwen3.8-27b davidau q4k_m max mtp. Made the panel from scratch over a week, ba
Top 3 papers on HF Daily Paper are all unusually delightful and interesting reads for anyone on the leading edge of local LLMs, agent harness optimization, etc, felt like sharing. DeepSeek-V4.1-Flash:
https://preview.redd.it/j13iin0jsfqh1.png?width=1129&format=png&auto=webp&s=3efdd2ad5c10de908aed1a673a6c7e6440177c09 https://preview.redd.it/k3fsrdspzfqh1.png?width=391&format=png&auto=webp&s=1c817687
Okay, this one seems probably really odd (or maybe not), but most harnesses operate within token spaces. That is, every tool output, even batching of tool outputs, go back up into inference calls. Lea
I was somewhat disappointed with the performance of Qwen 3.8 Flash Next on my single RTX 5090 using llama.cpp. One issue is that llama.cpp still has no Expert Caching implemented for MoE models. There
slowdown? what slowdown? Sources: Anthropic considers releasing a new AI model to counter OpenAI's momentum since Astra's launch, ahead of an IPO and after Amodei's call for a slowdown (Reuters) (Visi
6 billion tokens for $65 Is it as good as Astra and Fable at hard stuff? No of course not. But you don't always need a genius and can just call one for advice when you do. Extrapolate where this is go
arXiv:2609.20437v1 Announce Type: cross Abstract: This article presents a mathematical model of the Motivated Emotional Mind cognitive architecture developed for embodied intelligent systems. Such a s
arXiv:2609.19555v1 Announce Type: cross Abstract: Artificial intelligence for plant disease analysis has advanced from task-specific classifiers to multi-modal models capable of jointly interpreting v
arXiv:2505.17640v3 Announce Type: replace Abstract: Time series segmentation assigns a label to each part of a sequence. We formulate dense univariate segmentation as node classification on a graph wh
arXiv:2609.20080v1 Announce Type: new Abstract: The growing complexity of multi-domain operational environments (land, aerospace, naval, cyber, and electromagnetic spectrum) has increased the volume a
arXiv:2609.19524v1 Announce Type: new Abstract: Benchmark scores alone provide an incomplete basis for assessing the trustworthiness of modern artificial intelligence systems. Large language models (L
arXiv:2609.20249v1 Announce Type: new Abstract: Deep knowledge tracing (DKT) models implicitly decide which students an adaptive system believes have mastered a skill, yet almost all evidence on their
arXiv:2609.20130v1 Announce Type: cross Abstract: Recent memory-augmented repository-level program repair methods reuse historical repair experiences to improve LLM-based issue resolution. However, ou
arXiv:2602.03128v2 Announce Type: replace Abstract: Multi-agent LLM frameworks are data-intensive systems that govern how agents orchestrate tasks, manage state, and coordinate decisions. These archit
arXiv:2609.20034v1 Announce Type: cross Abstract: We present Astronex-World 1.0, an open controllable video world-model foundation. Given a text prompt (text-to-video) or an initial observation (image
arXiv:2512.13142v5 Announce Type: replace Abstract: People use LLMs for reproductive-health questions, including abortion-related support. A response can sound supportive while answers reinforce harmf
arXiv:2609.19989v1 Announce Type: new Abstract: The widespread adoption of LLMs has led to escalating content compliance risks. Prior works have contributed to addressing these risks in the English co
bonsai claim 98.2% intelligent retained, but their own documents show Ternary Bonsai 2 27B reaches 52.8 and 60.8, respectively, compared with 69.7 and 80.6 for Qwen3.5-27B, retaining roughly three qua
Build your own harness, folks. This is absolute banger paper from NVIDIA on self-evolving agent harnesses. (bookmark it) They introduce SoL-Pi which cuts token traffic by nearly half. And it matches i
arXiv:2506.17277v2 Announce Type: replace-cross Abstract: The retrieval stage of retrieval-augmented generation (RAG) for scientific question answering depends on how documents are segmented and how c
arXiv:2609.19425v1 Announce Type: new Abstract: Tool-augmented large language model (LLM) agents fail in a way no tool-selection or tool-security method addresses: they call tools that do not exist an
So I know that Ollama is built on top of llama.cpp, I decided to run a model on both and applied settings on llama.cpp like flash attention and q8/q4 kv cache. On Ollama (using the Python package) I w
arXiv:2609.20715v1 Announce Type: cross Abstract: Agent trajectories record what an agent does and what happens next. Yet standard supervised fine-tuning (SFT) applies loss only to agent-authored acti
Note not my work, but something i found and wanted to share so hopefully more people can push this along even further. https://github.com/nasone32/llama.cpp-RDNA3-7900xtx-opt I basically run 2 x 7900
arXiv:2609.19523v1 Announce Type: new Abstract: Web agents often revisit the same sites, yet most evaluations discard the procedures learned in earlier successful interactions. We introduce EconSkills
Engram Embedding Entendre is a model‑architecture optimization that extends standard token embeddings with learned multi‑token lookups, enabling recurring local patterns to be retrieved directly from
arXiv:2609.19375v1 Announce Type: new Abstract: Safety and comfort in human-robot physical in-teractions are essential aspects in the development of social technologies, where natural gestures, such a
arXiv:2609.20175v1 Announce Type: cross Abstract: The filter bubble is a notorious issue in Recommender Systems (RSs), which describes the phenomenon whereby users are exposed to a limited and narrow
arXiv:2609.06025v2 Announce Type: replace Abstract: Given various grammatical options, language users prefer the word order choice that reduces the overall length of syntactic dependencies, a principl
arXiv:2609.19596v1 Announce Type: new Abstract: Full-duplex speech models listen and speak at once, promising always-on assistants. Yet they must also decide when they should speak. Human listeners sp
arXiv:2609.20116v1 Announce Type: new Abstract: We study GPT-6-Astra in a zero-shot Vision-and-Language Navigation in Continuous Environments (VLN-CE) system, where it interprets instructions, assesse
arXiv:2609.20694v1 Announce Type: new Abstract: Mission-level autonomy for disaster response, search and rescue, and tactical UGV operations requires repeated path and mission planning as terrain cond
Hi everyone, The NeurIPS results aren’t out yet, but I’m expecting my paper to be rejected, so I’m considering submitting it to a journal instead. My scores were 2/3/3 (3/4/4). I feel like I would pro
arXiv:2609.19693v1 Announce Type: new Abstract: The rapid increase of deepfakes has raised significant concerns due to their spread on social media. Traditional multi-face forgery detectors crop and v
do you want some omni? here is omni for you 1. 🧭 Overview This repository hosts two checkpoints of the Realtime-Venus system: Realtime-Venus-Omni (Realtime-Venus-Omni/): the 9B audio-visual interactio
Introducing the first in our Parsed by LlamaParse series. We're outlining the stakes of 'getting parsing wrong' in consequential documents. We’re starting with the U.S. Energy Information Administrati
arXiv:2609.20218v1 Announce Type: cross Abstract: Large language models can label a tabular row from a plain-English description with no training - a capability now shipping in mainstream spreadsheet
arXiv:2609.20277v1 Announce Type: new Abstract: World Action Models (WAMs) have demonstrated strong robotic manipulation capabilities by augmenting pretrained video generative models with action exper
arXiv:2609.19717v1 Announce Type: cross Abstract: Large Language Models (LLMs) have achieved remarkable reasoning capabilities by utilizing chain-of-thought (CoT) as a scratchpad for intermediate stag
arXiv:2609.20252v1 Announce Type: cross Abstract: High-quality representations are essential for a wide range of downstream tasks. Dedicated embedding models are explicitly optimized for representatio
arXiv:2509.03405v2 Announce Type: replace Abstract: Language models (LMs) increasingly drive real-world applications that require world knowledge. However, the internal processes through which models
🚀 Meet Qwen3.8-Omni-Flash, Qwen's first omni-modal model built around agentic capabilities! Native audio-video understanding, reasoning, and tool use come together in one model: understand the content
arXiv:2609.19207v1 Announce Type: cross Abstract: Autoregressive transformer decoding is constrained by irregular key-value (KV) cache movement on tiled accelerators. Prior compression and DRAM-placem
arXiv:2609.20377v1 Announce Type: new Abstract: Autonomous driving involves coupled decision-making and scene evolution under multi-mode uncertainty. To capture this coupling and uncertainty, we intro