Knowledge catalogue
EXL3 is an alternative to llama.cpp. And while there is extensive tooling for llama.cpp, EXL3's primary deployment (TabbyAPI), has a OpenAI compatible API so it shouldn't matter. Why won't this tool m
arXiv:2603.11799v2 Announce Type: replace Abstract: Membership inference attacks (MIAs) are becoming standard tools for auditing the privacy of machine learning models. The leading attacks -- LiRA (Ca
arXiv:2608.14112v1 Announce Type: cross Abstract: Scientific simulations often produce scalar volumes faster than they can be stored, transferred, and loaded, while in situ reduction must use only a l
arXiv:2608.14049v1 Announce Type: new Abstract: Robotic manipulation of flat objects is challenging due to the ungraspable configurations and strong variations in object geometry and material. Existin
arXiv:2608.14106v1 Announce Type: cross Abstract: When forecasting hourly returns for 1,000 US equities, we observe an unexpected phenomenon: predictions become nearly flat and show poor stock ranking
arXiv:2608.13675v1 Announce Type: cross Abstract: Between October 2018 and July 2026 AI models progressed from simple systems like BERT to massive agents that solve complex math and write software. Th
arXiv:2608.13787v1 Announce Type: new Abstract: AI agents increasingly act on their users' behalf, handling tasks such as scheduling meetings, comparing offers, and haggling over prices. These princip
Structured outputs, web search, web fetch, MCP connector, and tool search are now available for Claude models hosted on Azure in Microsoft Foundry, turning a model endpoint into a production agent pla
arXiv:2608.14389v1 Announce Type: cross Abstract: Existing palm presentation attack detection (PAD) datasets are often limited by static imagery, restricted acquisition conditions, or insufficient mul
arXiv:2608.14496v1 Announce Type: cross Abstract: Cross-Tabular Data Generation (CTDG) seeks to learn a generative model from multiple heterogeneous tables and produce new synthetic tabular datasets.
Gemini and Pixel have entered long‑term partnership agreements with Arsenal FC, FC Barcelona, FC Bayern München, Liverpool FC, and Paris Saint‑Germain to deliver AI‑powered personal intelligence exper
Grok 4.6 is smart, super fast & affordable Grok 4.6 just broke into the Top 3 on the Artificial Analysis Healthcare & Medical Index Grok 4.6 (high) scores 50 - just ONE point off the top score of 51 I
Grok is very good at agentic tasks! Agentic work is where @grok 4.6 lands hardest, taking the top spot on the Artificial Analysis Agentic Index at 59, tied with Claude Opus 5 Max. The index measures t
arXiv:2608.14055v1 Announce Type: new Abstract: Authoritative scientific knowledge in geoscience remains largely trapped in legacy monographs and historical literature, where unstructured text and com
arXiv:2608.13922v1 Announce Type: new Abstract: Detecting distributional changes in high dimension is difficult when neither the pre-change nor post-change density is parametrically specified. We intr
Do you think today’s GPU-poor systems running Qwen 3.5 9B can outperform the ones we had with ChatGPT-4o? I mean, when 4o disappeared, people felt like they’d lost a great model, and now Qwen 3.5 9B Q
Trying to get a feel for where I stand. If you can list your relevant hardware and model used, that would be awesome. Here's mine: Model: Qwen3.8-27B-heretic-ara, Q5_K_M GGUF T/s: ~30-32 t/sec (I thin
Original Article - https://x.com/p_nawrot/status/2089315591010079034 I've spent the last few years working on efficient attention and KV Cache Compression. I've read many papers, dug deep into referen
I created 3 SVG prompts, each one rather hard. Perspective: A animated drone view perspective on a park. Beauty: A beach scene with an evil cat Composition: An AGI breaking out of a virtual sandbox pr
I'm back. Yesterday I released the first version of hyper-optimized Qwen3.8-27B inference engine for a RTX 3090, reaching 82 tps on single request and 672 peak. Over the last 24 hours I've been explor
IIRC, this is the 3rd month in a row that @FireworksAI_HQ is being featured on @tryramp's top software vendor list. This isn't just for inference / model serving (of which we serve 40T+ tokens a day),
arXiv:2608.13958v1 Announce Type: new Abstract: How do we govern AI systems whose reasoning we cannot fully inspect? Governance does not require understanding a system's reasoning. It requires stating
arXiv:2512.14338v4 Announce Type: replace Abstract: Many learning problems are organized by group symmetries. While invariance is often imposed through architectures or group averaging, we ask when it
Its interesting to see the experiments on how to give AI a computer: Codex & Claude Code use your local machine, ChatGPT Work on the web (as well as Claude and ChatGPT) give the AI a one-time machine
arXiv:2608.13580v1 Announce Type: cross Abstract: Jais 2 is a family of Arabic-centric large language models developed jointly by MBZUAI, Cerebras, and Inception, designed to advance Arabic-centric la
arXiv:2608.14425v1 Announce Type: new Abstract: LLM evaluations often use fixed sampling budgets, testing every item the same number of times even after estimates are precise. We introduce optstop, a
arXiv:2608.14191v1 Announce Type: cross Abstract: The key-value (KV) cache stores information from past tokens and is a major memory bottleneck in long-context inference. Existing quantization methods
arXiv:2608.14511v1 Announce Type: cross Abstract: High-order multiple-input multiple-output (MIMO) detection requires efficient search over a large discrete symbol space while producing reliable soft
arXiv:2608.14178v1 Announce Type: new Abstract: Tea is known as an important crop in many parts of South and Southeast Asia, yet the production of tea is still hampered by the multiple diseases that d
Support for the new ling 3.0 models has been merged into llama.cpp: https://github.com/ggml-org/llama.cpp/pull/26608#event-29549472828 Ling tiny 8b1b - https://huggingface.co/inclusionAI/Ling-3.0-tiny
arXiv:2503.01236v3 Announce Type: replace-cross Abstract: This paper addresses fixed-graph terrain-aware path refinement, in which a global planner is restricted to a predefined route space and may re
In medium reasoning mode, it both scores higher than the 3.6 version, AND is very much more efficient (almost half requests needed, and a third less tokens generated) - at DeepSeek v4 Flash 3107 MXFP4
arXiv:2608.14361v1 Announce Type: new Abstract: Intrinsic dimensionality (ID) is widely used to probe the representational complexity of language models, but it remains unclear whether ID differences
Hi all! I finally just got around to testing out Qwen 3.8 27b. I'm using Unsloth's UD-Q8_K_XL quant as a sit-in replacement to Qwen 3.6 27b, same quant size. Wow -- this thing isn't messing around. I
arXiv:2608.14068v1 Announce Type: cross Abstract: Conversational recommendation for e-commerce is increasingly mediated by large language models (LLMs), yet many real-world deployments operate under a
arXiv:2608.14282v1 Announce Type: new Abstract: Monocular temporal 3D detection aims to detect objects in 3D, given a monocular video. Query-based 3D detectors unify detection and cross-view associati
arXiv:2608.14074v1 Announce Type: new Abstract: AI agents increasingly act on external systems through standardized tool-calling protocols such as the Model Context Protocol (MCP), yet no infrastructu
Many people’s blindness to this problem — AI failing to follow instructions — is like their initially blasé attitude towards hallucinations, with the same fantasy that core, endemic problems would rap
arXiv:2608.14530v1 Announce Type: cross Abstract: Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose,
arXiv:2608.14015v1 Announce Type: cross Abstract: Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a pr
arXiv:2608.13883v1 Announce Type: new Abstract: Most agent-memory benchmarks test post-hoc recall, whereas MemoryArena evaluates whether memory supports interdependent, multi-session task completion.
for those looking for something small AND powerful, there is a new 1B (they claim, it looks more like 1.7B ...) model that claims to beat qwen 3.5 0.8B & 2B and gemma 4 E2B on a range of benchmarks. t
arXiv:2608.13606v1 Announce Type: new Abstract: The next generation of AI agents is increasingly moving beyond systems that answer isolated questions toward persistent personal assistants that can und
arXiv:2608.14054v1 Announce Type: new Abstract: Time series forecasting with pretrained foundation models has demonstrated strong zero-shot capabilities. However, achieving optimal performance on time
arXiv:2608.14264v1 Announce Type: cross Abstract: Model merging combines trained models directly in weight space, offering a compute-efficient alternative to additional fine-tuning. Selecting merge pa
arXiv:2608.13987v1 Announce Type: new Abstract: Nanbeige4.2-3B is a 3B-parameter agentic model built around a Looped Transformer (LT) that reuses one stack of layers for a second forward pass, adding
arXiv:2602.12498v2 Announce Type: replace Abstract: Negation is a fundamental linguistic operation in clinical reporting, yet vision-language models (VLMs) frequently fail to distinguish affirmative f
arXiv:2602.15572v3 Announce Type: replace Abstract: Agent-based modelling (ABM) is a widespread approach to simulate complex systems. Advancements in computational processing and storage have facilita
arXiv:2608.14349v1 Announce Type: new Abstract: We present a training-free method for multi-modal trajectory prediction that achieves comparable accuracy to a 57M-parameter transformer while requiring
NVIDIA Nemotron 3.5 Lightning, an open model built for high-volume agentic workloads, is now available in Amazon SageMaker JumpStart. This post shows how to deploy the 30B Mixture-of-Experts model (3B
Now that OpenCode plan isn't really worth it anymore, I'm looking for other options. I'm trying to figure out the Ollama Cloud subscription, specifically the $20 tier, because that's my budget for mon
Ollama has the best performance for deepseek v4 flash on average. For local only, you can try qwen3.8 that is optimized: Apple Silicon: ollama run qwen3.8:27b-mlx NVIDIA: ollama run qwen3.8:27b Benchm
I run locally Ollama + Qwen (tried multiples versions) and I connected it with MCP into my local SQLite database. When I ask a question, 90% of times I see it thinking and just stop and never answer t
arXiv:2608.14262v1 Announce Type: new Abstract: Temporal vision-language models (TVLMs) offer a reusable, prompt-based interface for surgical video understanding, yet, their robustness under clinicall
arXiv:2608.13923v1 Announce Type: new Abstract: Open-vocabulary 3D scene graphs provide compact semantic memory for language-guided navigation, but mapped objects are often exposed through a single fu
arXiv:2410.15658v4 Announce Type: replace-cross Abstract: Deep neural networks frequently produce overconfident, miscalibrated predictions. In ordinal classification, predictions must also adhere to a
arXiv:2608.14078v1 Announce Type: new Abstract: Localized 3D stylization aims to modify the appearance of a specified object part while preserving the remaining surfaces. In large reconstruction model
arXiv:2608.14094v1 Announce Type: cross Abstract: Cloud-local LLM inference systems have the potential to use the reasoning capability of large cloud models while protecting sensitive user data on per
arXiv:2608.14441v1 Announce Type: new Abstract: Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions a
pagedMark removes AI provenance from content you generated yourself. Two different things, and it is worth separating them. The first is metadata: C2PA Content Credentials, EXIF, XMP, IPTC, the genera