Knowledge catalogue
arXiv:2604.07747v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) can improve low-k reasoning accuracy while narrowing solution coverage on challenging math que
arXiv:2509.23322v2 Announce Type: replace Abstract: With the continuous expansion of Large Language Models (LLMs) and advances in reinforcement learning, LLMs have demonstrated exceptional reasoning c
arXiv:2604.06798v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) based large language models (LLMs) offer strong performance but suffer from high memory and computation costs. Weight binariz
arXiv:2411.19121v2 Announce Type: replace-cross Abstract: While text-to-video diffusion models have advanced significantly, creating coherent long-form content remains unreliable due to stochastic sam
arXiv:2604.03336v2 Announce Type: replace Abstract: BitNet b1.58 (Ma et al., 2024) demonstrates that large language models can operate entirely on ternary weights {-1, 0, +1}, yet no native binary wir
One great outcome of PaperWiki is personalized surveys. Survey papers continue to be one of the best ways to track a field. My agents are now generating personalized surveys on topics using my paper L
arXiv:2604.07788v1 Announce Type: cross Abstract: We introduce PeReGrINE, a benchmark and evaluation framework for personalized review generation grounded in graph-structured user--item evidence. PeRe
arXiv:2604.08499v1 Announce Type: cross Abstract: Prompt injection attacks pose serious security risks across a wide range of real-world applications. While receiving increasing attention, the communi
arXiv:2604.06228v1 Announce Type: cross Abstract: We introduce probabilistic language tries (PLTs), a unified representation that makes explicit the prefix structure implicitly defined by any generati
arXiv:2505.24499v2 Announce Type: replace Abstract: Generating high-quality Scalable Vector Graphics (SVGs) is challenging for Large Language Models (LLMs), as it requires advanced reasoning for struc
arXiv:2604.06811v1 Announce Type: cross Abstract: Skill-based agent systems tackle complex tasks by composing reusable skills, improving modularity and scalability while introducing a largely unexamin
arXiv:2603.04759v2 Announce Type: replace Abstract: The limited context window of contemporary large language models (LLMs) remains a primary bottleneck for their broader application across diverse do
arXiv:2604.07415v1 Announce Type: cross Abstract: Large language models (LLMs) are probabilistic in nature and perform more reliably when augmented with external information. As complex queries often
arXiv:2604.07101v1 Announce Type: cross Abstract: We present the Surveillance Forgery Image Test Range (SurFITR), a dataset for surveillance-style image forgery detection and localisation, in response
arXiv:2604.07513v1 Announce Type: cross Abstract: AI-based persona simulation -- often referred to as digital twin simulation -- is increasingly used for market research, recommender systems, and soci
arXiv:2509.09794v4 Announce Type: replace Abstract: Computational models have emerged as powerful tools for multi-scale energy modeling research at the building and urban scale, supporting data-driven
The challenge with agent harnesses is not only building them but distribution. If we want agent products to scale like saas, the interfaces and ecosystem still need a fundamental breakthrough. im exci
The Information wrote that Meta consumes 60.2T tokens per month, or ~750M per employee. Meta is getting absolutely token-mogged by us currently. SemiAnalysis employees consume 1.86B tokens per month.
The power of JAX Introducing gyaradax 🐉: A JAX solver for local flux-tube gyrokinetics with custom CUDA kernels for acceleration. This entire code was vibecoded by @ggalletti_ and me in a month. Valid
arXiv:2604.06621v1 Announce Type: cross Abstract: Dr. David Blackwell was a mathematician and statistician of the first rank, whose contributions to statistical theory, game theory, and decision theor
When using Ollama Pro's cloud models with OpenClaw, a local Ollama installation is not strictly required — OpenClaw is an AI agent execution layer that handles tools, memory, scheduling, and messa...
arXiv:2604.06205v1 Announce Type: cross Abstract: The growth of online platforms and user content requires strong content moderation systems that can handle complex inputs from various media types. Wh
arXiv:2510.12476v2 Announce Type: replace Abstract: Large language models (LLMs) have grown more powerful in language generation, producing fluent text and even imitating personal style. Yet, this abi
arXiv:2604.08281v1 Announce Type: new Abstract: Large reasoning models (LRMs) have achieved strong performance enhancement through scaling test time computation, but due to the inherent limitations of
arXiv:2604.08032v1 Announce Type: cross Abstract: Automated maritime collision avoidance will rely on human supervision for the foreseeable future. This necessitates transparency into how the system p
wow. GenAI is definitely not living up to the hype, for most people. 🦔A global survey of 3,750 executives and employees found that 54% of workers bypassed their company's AI tools in the past 30 days
Wrapped up the 3-day AI Engineer conference in London. Key takeaways: Planning & verification > implementation. Humans define what to build and validate it works. AI handles the coding. The harness ma
(1/5) FP4 hardware is here, but 4-bit attention still kills model quality, blocking true end-to-end FP4 serving. To fix that, we propose Attn-QAT, the first systematic study of quantization-aware trai
Editor’s note: This blog post outlines Google Cloud’s GPU AI/ML infrastructure reliability strategy, and will be updated with links to new community articles as they appear. As we enter the era of mul
*AI Systems Performance Engineering* by Chris Fregly is a ~1,000-page book (published December 2025) covering GPU/CUDA kernel tuning, PyTorch optimization, distributed training, and high-throughput...
llama.cpp release **b8737** is a focused maintenance build that adds missing CUDA error handling to the ggml backend. Specifically, it checks the return values of NVIDIA CUB library calls used in t...
Compliance workflow automation startup Haast Pty Ltd. said today it has raised $12 million in an early-stage round of funding to help organizations accelerate the rollout of artificial intelligence-ge
I was unable to retrieve the specific X (Twitter) post at the URL provided, as the content is not accessible through search results. The tweet's text snippet ('memory is context, and context is wha...
🎙️Introducing Max Agency Max Agency is a new podcast where we go deep on how the best agents are actually being built: architecture decisions, tradeoffs, evals, and everything in between. Each episode
NEW paper from Microsoft Every agent benchmark has the same hidden problem: how do you know the agent actually succeeded? Microsoft researchers introduce the Universal Verifier, which discusses lesson
Nvidia published DWDP (Distributed Weight-Data Parallelism), a new inference parallelism strategy focused on prefill. It sounds slightly insane until you remember the target machine is GB200 NVL72. Th
Amazon released the Nova 2 model family on December 2, 2025 at AWS re:Invent, comprising Nova 2 Lite, Nova 2 Pro (Preview), Nova 2 Omni, and Nova 2 Sonic, all featuring a 1-million-token context wi...
Hermes Agent is an open-source, self-improving AI agent developed by Nous Research that features a built-in learning loop — it creates skills from experience, builds persistent memory across sessio...
This 10-min read from @Vtrivedy10 changes how you build AI agents. Most people are stuck in the same loop; switching models when agents break. The reframe: evals are the training data for your harness
🤯 AI doesn’t need to be AGI to cause harm! 🤯 ChatGPT can’t reliably run a timer but it has still been implicated in delusions, suicides, cognitive surrender, mass disinformation, and so much more. A s
In different hands, Mythos would be an unprecedented cyberweapon I am not sure how we deal with this, except to note a narrow window where we know only 3 companies could be at this level of capability
NVIDIA has introduced a modular, library-based architecture for Omniverse, exposing core components—RTX rendering, PhysX-based simulation, and data storage pipelines—as standalone, headless-first C...
It is easy to be fooled that there is meaning here, but that is because you are trained to provide the meaning. Think of the characters. Think of the emotional arc. Think of the unsaid stuff you need
NEW paper from Google on multi-agent research agents. It's one of the first systems that handles end-to-end LaTeX generation, targeted literature reviews, and conceptual diagrams as a decoupled, stand
Vultr has upgraded its Identity and Access Management (IAM) framework by replacing legacy ACLs with fine-grained, policy-driven access controls, enabling organizations to assign roles, manage permi...
On Mythos: I know it feels utterly wrong and bad to say it's better/safer if this capability was widely available and e.g. open sourced, but it actually might be. Anthropic could now be sitting on so
🏆 Qwen3.6-Plus just swept OpenRouter’s Daily, Weekly & Trending charts! The trial has wrapped. The model is fully live & production-ready. 📉 Lower latency | 🔥 Top-tier reasoning | 💰 Unmatched $/token
Some approaches: 1) Multiple reviews. Some papers already show having many AI team members review a problem reduces errors 2) Building in tests and checkpoints 3) Multiple independent answers that are
.@steipete has been an amazing supporter of open models and local models. This is even before the rise of OpenClaw. 🦞 Ollama is here to love. ❤️ Let's make the world a better place together. Some folk
I was unable to directly access the specific X (Twitter) post or the underlying paper it references, and my general web search did not surface the specific study being discussed. I cannot responsib...
We're seeing even more autonomous AI coworkers. The new MLE agent on the market is Disarray. In Kaggle competitions, Disarray: - won 28 medals across diverse domains (vision, NLP, tabular data) - plac
Anthropic didn't release their latest model, Claude Mythos (system card PDF), today. They have instead made it available to a very restricted set of preview partners under their newly announced Projec
GLM 5.1 is live on Fireworks! SOTA for agents and coding: →Plans and executes multi-hour workflows without falling apart →Planning, executing, testing, and refining over hundreds of rounds to deliver
How can you improve your agentic search pipeline? I just wrote a blog post with @tech_optimist from @lancedb to answer exactly that. TLDR: - Parse files and take page-level screenshots with LiteParse,
I was told about the Mythos release, but didn't have access, so have no personal experience to add. Two points from brief: 1) It is not built for IT security, it is just a good enough model that it is
More on the @nytimes piece about that $1.8B, two-person, AI company ... not the paper's finest moment in quick retrospect. And in the healthcare context no less. Our friend @GaryMarcus was on this a f
This is a great tutorial (credits @itsclelia + @lancedb) on how to build a practical retrieval pipeline that integrates directly with your agent harness. 1. Ingest a massive pile of docs with litepars
tldr: everyone is converging on the same product shape: a general harness that takes a goal, uses tools, and does knowledge work. once every product is a harness, the next frontier is the feedback loo
Visually rich documents are especially challenging for agents. Tables, charts, and images often break traditional document pipelines, making complex reasoning difficult📄 So we teamed up with @lancedb
Vercel’s unified API for hundreds of models — one endpoint, auth, failover, spend controls; pairs with the AI SDK and powers this wiki’s Ask models.