Knowledge catalogue
releasing qwen3.6 35b a3b carnice edition (inspired by @kaiostephens) to enjoy an hermes agent tailored finetuned on unified memory hardware! safetensors and gguf versions (q8, 6,5 and 4) available! f
Simon Willison documents the differences between Anthropic's Claude Opus 4.6 and 4.7 system prompts, analyzing changes that Anthropic made public. The notes likely highlight modifications to model beh
Source tweet is here: https://x.com/jerryjliu0/status/2044902620746363016?s=20 We comprehensively benchmarked Opus 4.7 on document understanding. We evaluated it through ParseBench - our comprehensive
Mark Gurman / Bloomberg: Sources: the glowing “26” in Apple's WWDC invite is teasing a revamped Siri, memory shortages may push Mac Studio and touch MacBook Pro launches by a few months — Also: Memory
Thanks for the shoutout re: ParseBench! 📑 🙂 Opus 4.7 is definitely a step up from Opus 4.6 on document understanding capabilities. Great to see labs start to push a bit more on this front. [16 Apr 202
The biggest challenge of using chat-based AI systems is that the details of what they can do are invisible - those tool descriptions are the missing manual, publishing them would be a huge benefit to
The continuing gap between the capabilities of Gemini Pro 3.1 (very good model) and the capabilities of the Gemini app/website is odd. The model can do what Claude/GPT can do, but there is a minimal h
The Top AI Papers of the Week (April 13 - 19) - AlphaEval - AiScientist - Auto-Diagnose - Nemotron 3 Super - Subliminal Learning - Automated W2S Researcher - Memory Transfer Learning Read on for more:
Who is running local models on GPUs on OpenClaw? I have started benchmarking different models this week. I am working on improving model selection and switching UX on OpenClaw, i.e. I run /model vllm/
This post documents a Wordle game result where the player solved puzzle #1,764 in three attempts, using the standard color-coded emoji system (black for incorrect letters, yellow for correct letters i
This post explains the technical details behind common terms and tools encountered on Hugging Face for running large language models locally, including quantization formats (GGUF, Q_4_M, IQ_4XL), opti
You should watch this. It just shows how disconnected we are from the small group of people making decisions that will impact our future heavily. These people have so much ai psychosis. If you listen
A downside with using VLMs to parse PDFs is guaranteeing that the output text is *correct* and output in the correct reading order. 1️⃣ Text correctness: making sure that digits, words, sentences are
Agentic Engineering Patterns > Here's an example of a deceptively short prompt that got a lot of work done in a single shot. First, some background. I send out a free Substack newsletter around once a
Stephanie Stacey / Financial Times: Airbnb launches a pilot in NYC, LA, and other cities that lets users to select from a range of boutique hotels alongside private homes in a bid to boost growth — Co
Anthropic shut down an entire company's Claude access overnight 60+ employees. No explanation. Just an email. Want to appeal? Fill out a Google Form. Integrations gone. Histories gone. Everything buil
Anthropic’s CEO keeps talking about AI wiping out jobs because he’s trying to IPO this year. If he positions Claude as armageddon for jobs, his TAM becomes “all white-collar human labor,” not just AI
Both Gemini and Grok are underrated. I used to champion Gemini for a while, but lately I've been very happy with Grok, especially since the Grok 4.20 release with multi-agent. And now we have Grok 4.3
Research: Claude system prompts as a git timeline Anthropic publish the system prompts for Claude chat and make that page available as Markdown. I had Claude Code turn that page into separate files fo
Fantastic to see GLM being applied to such fresh, dynamic scenarios. Doing some stress tests on OpenMAIC’s Interactive Simulation with a DNA Replication case. 💻 Both powered by @Zai_org — with GLM-5.1
GDPval is one of the most important benchmarks of AI ability because it is based on human expertise. It compares expert human performance to AI performance using expert human judges who spend an avera
Hermes Agent is model & tool backend agnostic for a reason, everyone should have access to AI. We don't dictate the rules of use for your agent, YOU do Anthropic shut down an entire company's Claude a
I prefer my design tool to be more closely integrated with where my agents work. I spent a few hours building my own design tool (inspired by Claude Design) inside my orchestrator. I can use this with
I think Artificial Analysis does a good job overall and provides transparency in benchmarking, but GDPval-AA is not a good benchmark and needs to stop being reported. It is Gemini 3.1 judging other mo
Iain Martin / Forbes: Mistral, which once aimed for top open models, now leans on being an alternative to Chinese and US labs, says it's on track for $80M in monthly revenue by Dec. — Paris-based Mist
Okay this one is insane. A new 18B frankenstein model was just released on @huggingface — Beats the new Qwen3.6-35B-A3B on a 44-test suite despite requiring 12GB VRAM instead of 24GB 🤯 Runs on a SINGL
RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second. Nothing touches the cloud. The trick everyone else missed:
Michael Nuñez / VentureBeat: Salesforce announces Headless 360, an initiative that will give AI agents access to Salesforce's platform capabilities through APIs, MCP tools or CLI commands — Salesforce
Stunning views and just as jaw-dropping of presentations at the @cerebral_valley Gemma 4 Launch Party tonight! @vllm_project @UnslothAI @Ollama @huggingface @cactuscompute @apple @nvidia @pytorch and
We push Prefill/Decode disaggregation beyond a single cluster: cross-datacenter + heterogeneous hardware, unlocking the potential for significantly lower cost per token. This was previously blocked by
This post documents a Wordle game result where the player solved puzzle #1,763 in 5 attempts, with the final answer being a five-letter word. The emoji grid shows the progression of guesses, indicatin
30K+ likes in the first hour. 🤯 That is crazy! Design is unsolved with agents. But lots of impactful work generated by agents is around design. Claude Design is Anthropic's way of saying that they are
arXiv:2601.05991v2 Announce Type: replace Abstract: In safety-critical domains, linguistic ambiguity can have severe consequences; a vague command like 'Pass me the vial' in a surgical setting could l
arXiv:2604.14635v1 Announce Type: new Abstract: We present a curated multi-platform LiDAR reference dataset from an instrumented ICOS forest plot, explicitly designed to support calibration, benchmark
arXiv:2604.15238v1 Announce Type: cross Abstract: This paper investigates continuous-time and discrete-time firing-rate and Hopfield recurrent neural networks (RNNs), with applications in nonlinear co
arXiv:2511.15915v2 Announce Type: replace-cross Abstract: We present AccelOpt, a self-improving large language model (LLM) agentic system that autonomously optimizes kernels for emerging AI acclerator
arXiv:2604.14682v1 Announce Type: cross Abstract: Speculative decoding accelerates large language model (LLM) inference. It uses a small draft model to propose a tree of future tokens. A larger target
arXiv:2511.18107v2 Announce Type: replace Abstract: Accurately solving partial differential equations (PDEs) is critical to understanding complex scientific and engineering phenomena, yet traditional
arXiv:2604.15291v1 Announce Type: new Abstract: The reliability of a machine vision system for autonomous driving depends heavily on its training data distribution. When a vehicle encounters significa
arXiv:2604.14902v1 Announce Type: cross Abstract: Intelligent embodied agents should not simply follow instructions, as real-world environments often involve unexpected conditions and exceptions. Howe
arXiv:2604.14853v1 Announce Type: new Abstract: Test-time compute scaling, the practice of spending extra computation during inference via repeated sampling, search, or extended reasoning, has become
arXiv:2604.14655v1 Announce Type: cross Abstract: We present AgentGA, a framework that evolves autonomous code-generation runs by optimizing the agent seed: the task prompt plus optional parent archiv
arXiv:2512.13671v2 Announce Type: replace Abstract: Industrial anomaly detection (IAD) is challenging due to the subtle and highly localized nature of many defects, which single-pass vision--language
arXiv:2604.13940v1 Announce Type: new Abstract: Scientific peer review faces mounting strain as submission volumes surge, making it increasingly difficult to sustain review quality, consistency, and t
This article from Latent Space discusses Anthropic's Claude Opus 4.7 release, highlighting incremental improvements across multiple performance dimensions compared to the previous 4.6 version. The pie
arXiv:2604.15299v1 Announce Type: new Abstract: Video generation has advanced rapidly, with recent methods producing increasingly convincing animated results. However, existing benchmarks-largely desi
arXiv:2601.13503v2 Announce Type: replace Abstract: Psychiatric narratives encode patient identity not only through explicit identifiers but also through idiosyncratic life events embedded in their cl
The latest addition to Anthropic PBC’s product portfolio is Claude Design, a tool that enables users to generate visual assets with prompts. The company launched the offering into public preview today
The Trump administration has spent nearly two months fighting with AI company Anthropic. It's dubbed the company a 'RADICAL LEFT, WOKE COMPANY' full of 'Leftwing nut jobs' and a menace to national sec
arXiv:2604.13109v1 Announce Type: cross Abstract: We present a two-stage pipeline for AI-assisted improvement of published algorithm implementations. In the first stage, a large language model with re
arXiv:2603.23682v2 Announce Type: replace-cross Abstract: The rapid adoption of large language models (LLMs) in education raises profound challenges for assessment design. To adapt assessments to the
arXiv:2505.10846v3 Announce Type: replace Abstract: This paper presents AutoRAN, the first framework to automate the hijacking of internal safety reasoning in large reasoning models (LRMs). At its cor
arXiv:2604.14472v1 Announce Type: new Abstract: Physics-informed neural networks (PINNs) are often selected by a single scalar loss even when the quantity of interest is more specific. We study a hybr
arXiv:2604.15202v1 Announce Type: new Abstract: Coverage path planning on irregular hexagonal grids is relevant to maritime surveillance, search and rescue and environmental monitoring, yet classical
arXiv:2604.14171v1 Announce Type: new Abstract: Romanized Nepali, the Nepali language written in the Latin alphabet, is the dominant medium for informal digital communication in Nepal, yet it remains
arXiv:2604.15297v1 Announce Type: new Abstract: MLP is a heavily used backbone in modern deep learning (DL) architectures for supervised learning on tabular data, and AdamW is the go-to optimizer used
arXiv:2604.14860v1 Announce Type: cross Abstract: We study bandit best-arm identification with arbitrary and potentially adversarial rewards. A simple random uniform learner obtains the optimal rate o
arXiv:2604.14389v1 Announce Type: new Abstract: Automated fact-checking in dialogue involves multi-turn conversations where colloquial language is frequent yet understudied. To address this gap, we pr
arXiv:2411.05472v2 Announce Type: replace Abstract: The paradigm shift toward structure-driven molecule generation has been propelled by advances in deep generative models, such as variational auto-en
arXiv:2604.15088v1 Announce Type: new Abstract: Building extraction from optical Remote Sensing (RS) imagery suffers from performance degradation under real-world hazy and low-light conditions. Howeve