Knowledge catalogue
arXiv:2604.07223v1 Announce Type: cross Abstract: As large language models (LLMs) evolve from static chatbots into autonomous agents, the primary vulnerability surface shifts from final outputs to int
arXiv:2604.06387v1 Announce Type: cross Abstract: Accurate reconstruction of environmental scalar fields from sparse onboard observations is essential for autonomous vehicles engaged in aquatic monito
arXiv:2512.13040v2 Announce Type: replace-cross Abstract: Detecting fraud in financial transactions typically relies on tabular models that demand heavy feature engineering to handle high-dimensional
arXiv:2604.08522v1 Announce Type: new Abstract: Video temporal grounding (VTG) is typically tackled with dataset-specific models that transfer poorly across domains and query styles. Recent efforts to
arXiv:2604.06182v1 Announce Type: cross Abstract: Existing online benchmarks for mobile GUI agents remain largely app-centric and task-homogeneous, failing to reflect the diversity and instability of
arXiv:2509.08016v2 Announce Type: replace Abstract: Video Large Language Models (VideoLLMs) face a critical bottleneck: increasing the number of input frames to capture fine-grained temporal detail le
arXiv:2604.06277v1 Announce Type: new Abstract: Existing hallucination detection methods for large language models (LLMs) rely on external verification at inference time, requiring gold answers, retri
arXiv:2604.06367v1 Announce Type: cross Abstract: Web agents automate browser tasks, ranging from simple form completion to complex workflows like ordering groceries. While current benchmarks evaluate
arXiv:2604.08510v1 Announce Type: new Abstract: Large language models (LLMs) can perform remarkably complex tasks, yet the fine-grained details of how these capabilities emerge during pretraining rema
arXiv:2604.08513v1 Announce Type: new Abstract: Transfer learning followed by fine-tuning is widely adopted in medical image classification due to consistent gains in diagnostic performance. However,
arXiv:2604.07957v1 Announce Type: cross Abstract: Vision-language models (VLMs) and generative world models are opening new opportunities for embodied navigation. VLMs are increasingly used as direct
Yet another illustration of why LLMs aren’t even close to being AGI. The world’s best LLMs are still terrible at poker. We put each model into a 200bb heads-up NLHE match against GTO Wizard AI. The be
>8 out of 8 [cheap oss] models detected Mythos's flagship FreeBSD exploit Completely disingenuous They gave it just ~20 lines of code to read. They baked in custom, relevant context pertinent to the e
a useful mental model on how teams can think about good data design to improve their models/agents: Evals ~= Training Data ~= Environments - in Classical Deep Learning, we learn from each training exa
After playing with it a bit, Meta's Muse Spark Thinking is fine so far, but really doesn't match the current Big Three models. It also is a bit... weird. Like some strange language & tone, a little lo
Automating Loan Income Verification with AI agents 💸☑️ Loan processors spend a huge portion of time verifying the borrower’s documents (tax forms, pay stubs, bank deposits), and cross-checking them fo
The ICML 2026 'final justification' is a **new requirement this year** where reviewers must submit a written explanation of their final recommendation after reading author rebuttals. New for ICML ...
maybe some nuance 😄 I don’t think anyone is “lying” about how great Mythos will be —> but there’s expectation misalignment between the Test Harness set up for Mythos and a belief it was given this cra
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...
Tesla puts a lot of effort into ensuring that our cars don’t run over animals More than 350 million vertebrate animals are killed by human driven vehicles every year. In general, autonomous vehicles a
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
Users are running Google's Gemma 4 models locally via Ollama and accessing them on their phones through Open WebUI, reporting a positive experience. Since Ollama doesn't run natively on iOS or And...
Yesterday’s Mythos announcement from Anthropic was overblown. • Sandboxing was turned off, so test didn’t show much about the real world. • Cheap open-weight models can (already) do some similar stuff
Claude Mythos is Delusional Media Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software. It’s powered by our newest frontier model, Claude Mythos Previe
Curriculum Learning for harnesses - should we teach agents how we teach kids? start small and easy and progressively get harder for my research friends here's an under-explored area we're thinking abo
Director James Cameron on why Big Tech owning AGI is scarier than any science fiction he's ever made: 'AGI will not emerge from a government funded program. It will emerge from one of the tech giants
GLM-5.1 is Z.ai's post-training upgrade to GLM-5, now available on Together AI, delivering a 28% coding performance improvement through a refined reinforcement learning pipeline while retaining the...
Highlights: 👉 28% coding improvement over GLM-5 with refined RL post-training 👉 Better long-horizon execution across hundreds of rounds and thousands of tool calls 👉 Thinking mode, tool calling, and s
Introducing GLM-5.1 from @Zai_org on Together AI. AI natives can now use GLM-5.1 on Together and benefit from reliable inference for production-scale agentic engineering and long-horizon coding workfl
Plus: execute_code on remote backends, Firecrawl cloud browser, Browser Use provider, xAI prompt caching, thinking prefill continuation, cron script injection, Telegram reactions, shared thread sessio
The EU deep state is telling us (via state-owned media and Soros-funded NGOs) that Telegram is a PROBLEM because people can discuss content from OTHER social media in PRIVATE telegram groups. This non
The new model from Meta is already looking like a disappointment: overoptimized for public benchmark numbers at the detriment of everything else. Knowing how to evaluate models in a way that correlate
To anyone who read Rebooting AI back in (checks notes) 2019, this is both hilarious and unsuprising. The field has wasted 7 years on an architecture that can’t solve one of the most basic litmus tests
Nous Research's **Hermes Agent** is an open-source agentic framework by NousResearch that delivers significantly improved reliability when using GPT-5.x or Codex (OpenAI Codex subscription) models ...
I was unable to retrieve the specific content of the linked X (Twitter) post from Anthropic at that URL, and the web search did not return results directly related to it. The post ID (2042020348451...
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
**WorldSim** by Nous Research (worldsim.nousresearch.com) is an AI-powered interactive simulation platform that lets users explore open-ended, AI-generated worlds and environments. It includes mult...
Z.AI's GLM-5.1, a next-generation flagship model for agentic engineering released in April 2026, is now available across all Kilo Code surfaces — including the VS Code extension, Cloud Agents, and ...
GLM-5.1 is Z.ai's (zai-org) next-generation open-source flagship model for agentic engineering, achieving state-of-the-art performance on SWE-Bench Pro and significantly outperforming its predecess...
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,
INCREDIBLE GLM-5.1 weights are now opensource > i’ve had early access to the weights for the past few days > and yeah… this one matters a lot benchmarks? > SWE-Bench Pro: 58.4 > beats Opus 4.6 (57.3)
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
SuperClaude (Mythos) still seems irreducibly Claude-y given the transcripts in the system card. Here two versions of Mythos are forced to talk to each other across multiple rounds. They are less philo
You have heard of @openclaw competitor from @NousResearch called “Hermes.” Tomorrow at 4 pm we will get nerdy with @theemozilla. Live. I will get people up who asks questions here first. https://x.com