This should be impossible! With only 3 lines, this open-source library finds every bad image in your dataset: • Blurry • Under/over-exposed • Oddly sized • Duplicates If you are into AI/ML, star this repo: github.com/cleanlab/clean…. They are building incredible things!

At today’s White House meeting on AI: - OpenAI - people who left OpenAI because it wasn’t focused enough on existential risk - people who bought the exclusive rights to everything OpenAI makes - Google

this is what it feels like

i have noticed that LLMs like claude and gpt-4o work really well with this prompt, it instructs them to 'contemplate' for a bit before giving the final answer.

Just visited the men’s washroom at Indore Airport and was impressed by how spotless it was. A big shoutout to Shreeram ji, the cleaner, for his hard work and dedication. Such individuals truly make a difference. Kudos to him! Airports Authority of India Swachh Survekshan

DeepSeek r1 is exciting but misses OpenAI’s test-time scaling plot and needs lots of data. We introduce s1 reproducing o1-preview scaling & performance with just 1K samples & a simple test-time intervention. 📜arxiv.org/abs/2501.19393

My team is hiring researchers! I’m primarily interested in candidates who have (i) several years of experience doing excellent work as a SWE or RE, (ii) who have substantial research experience of some form, and (iii) who are familiar with modern ML and the AGI alignment

Announcing AI Dev 25: A conference for AI developers, this Pi day (3/14/2025)! There're great AI academic conferences for researchers (NeurIPS, ICLR, ICML, etc.) and some companies hold great meetings around their products (Google I/O, OpenAI DevDay, etc.). But we need more

Chain-of-Associated-Thoughts (CoAT) is a new framework that enhances LLMs' reasoning abilities by combining Monte Carlo Tree Search with dynamic knowledge integration. The framework addresses the limitations of existing "fast thinking" approaches by introducing an "associative

Very succinct

Byte Pair Encoding

A new feature from Anthropic: "Think" tool for Claude Once Claude starts generating a response, "think" tool creates dedicated space to pause and reflect on whether Claude has gathered all necessary information. It's not the same feature as “extended thinking,” which happens

Blog post: anthropic.com/engineering/cl…

Neat report from Microsoft providing a taxonomy of failure modes in agentic AI systems. If you are building agentic systems today, you will run into many issues. Some of the common ones are summarized in this report. Great resource for AI devs.

Beyond the Myriad of Models: A Systems Approach to Real-World ML open.substack.com/pub/shivamnegi…

Just finished reading Chip Huyen's "ML System Design". Highly recommend it to anyone looking to deepen their understanding of system design in the ML space! 📚🤖 #MLSystemDesign #ChipHuyen #MachineLearning open.substack.com/pub/shivamnegi…

I didn't want to post on Grok safety since I work at a competitor, but it's not about competition. I appreciate the scientists and engineers at xAI but the way safety was handled is completely irresponsible. Thread below.

A #computervision researcher at #ICML2025

Another massive week of AI and robotics news. I summarized everything from OpenAI, xAI, Google, Cognition, Figure, AWS, Meta, UBTECH, Pollen Robotics, Agility Robotics, and more. Here's everything you need to know and how to make sense out of it: