To wait for rate limits, claude just writes a python function to wait, pretty nice

HUGE CAVEAT: Ouf this is a lot worse than I thought and conveyed. It's based on a 120-problem split from the ARC-AGI-II split assuming pass@250 !!! That means it only needs to get one out of 250 correct and doesn't even need to know which one! That's much more lax than the

Trelis AI Grants Update - 2Q 2025 --- 1. Congratulations to Dima Yanovsky (Dima Yanovsky) for completing his grant - "Accelerating Robotics Imitation Learning via Simulation and AR Teleoperation." - Note: Grants are announced each quarter based on completion, so there will be some

o3 / o4-mini causing bad engineering practices Hard enough to motivate myself to inspect traces when they exist. Impossible when they don’t

Funny saying grok is AGI BUT def impressive from grok 4

Program Synthesis approach breakthrough in ARC-AGI through Self-Play 📖 Read 201: « Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on Arc-AGI », by Pourcel Julien @ICML, Cédric and Pierre-Yves Oudeyer github.com/flowersteam/SO… The work of the authors is a

Excited to say that Ronan McGovern of Trelis Research and I have teamed up to work on ARC-AGI-2! Here's my latest on framing and approach, and a summary thread below: lewish.io/posts/how-to-b… Or in video form: x.com/TrelisResearch…

🤖Train an ACT Policy for the SO-101 Robot🤖 --- This is the third video in the Trelis series on robotics! I describe how to collect data for training, and then I train the ACT policy for an SO-101 robot, using the LeRobot library from Hugging Face ! I then evaluate

New video by community members Trelis Research and Lewis Hemens on their ARC Prize 2025 progress * Their approach to solving ARC-AGI * A call for sponsors * Research plan x.com/TrelisResearch…

npx ccusage@latest (courtesy of Simon Willison's great blog). Shows claude code usage. I've been on the 90 EUR per month plan, seems heavy users will be loss making for Antropic here. I downgraded to the 20 per month because I'm not quite using it heavily (except if I need a big

.ARC Prize listed on the Qwen model card 2nd model card for us in 2 weeks Excited for ARC-AGI to be seen as a supported way to measure model performance x.com/Alibaba_Qwen/s…

Anyone have a connection at Qwen? Trying to reproduce the results on ARC Prize and getting different metrics Want to get a hold of them and find out how they tested

- This is a **pure neural transductive approach**. - **BUT**, it trains a planning type module that has **no direct access to the problem input-pairs**. - The net is trained only on **one input-output pair at a time**, but also takes in positional embeddings for the grids AND an

New ARC Prize 2025 High Score 21.6% by Giotto. ai (Aldo Podestà)

Launching GitArsenal CLI, a coding agent that can setup any github repository on your chosen hardware. > It will spin up your preferred GPUs and do the setup there. > It will debug the issues with agents and resolve all dependency conflicts. > It won't stop until it runs your

+ GPT OSS Release, Inference and Fine-tuning + - My notes on the OpenAI models - Inference with vLLM - Fine-tuning with Hugging Face transformers Full video on Youtube on the Trelis Research channel TIMESTAMPS: 00:00 Introduction to OpenAI's OSS Models 00:16 Model