In the past I've read the advice on writing good ML papers from Neel Nanda Andrej Karpathy Sebastian Farquhar and many others. So I thought: why not distill all of them into a Claude skill? Now I have an elite research writing partner at hand. Check it out for ICML! github.com/zechenzhangAGI…

I don't think people have realized how crazy the results are from this new TTT + RL paper from Stanford/Nvidia. Training an open source model, they - beat Deepmind AlphaEvolve, discovered new upper bound for Erdos's minimum overlap problem - Developed new A100 GPU kernels 2x

Excited to anounce the launch of the podcast - Hyperplane - that aiexplorations and I are doing. It will cover the entire breadth of AI: theoretical to tactical. Both aiexplorations and I have spent 20+ years building data and AI products. We've seen what happens when you

"Radiologists work in the most advanced digitized field in healthcare. Yet, why are they getting burned out? Why are they still doing manual geometry measurements?" In the latest episode of the Hyperplane podcast, aiexplorations and I demystify the "black box" of radiology. We

I'm Boris and I created Claude Code. I wanted to quickly share a few tips for using Claude Code, sourced directly from the Claude Code team. The way the team uses Claude is different than how I use it. Remember: there is no one right way to use Claude Code -- everyones' setup is

Software development is undergoing a renaissance in front of our eyes. If you haven't used the tools recently, you likely are underestimating what you're missing. Since December, there's been a step function improvement in what tools like Codex can do. Some great engineers at

Short thread on the first-ever strong image captioning model: BLIP In 2021 CLIP came out, showing the power of training on large scale image-caption pairs. CLIP is strong but limited, as it is not generative. [1/4]

This drop from Sarvam AI is really just so good I keep going back to it! Some of these fonts are so wild, it's magical they can even be transcribed. What's next team Sarvam AI?

BLIP will be remembered as a classic A goated image captioning model ahead of its time Still very popular on the hub with 2M downloads: huggingface.co/Salesforce/bli…

We are in the middle of exam season for children, and once March ends, there will be a flood of messages everywhere asking parents to enroll their kids into summer camps. Most of these camps focus on fairly standard things. Swimming, art and craft, some STEM based science

It is crazy that massive book piracy was so critical to the early days of both OpenAI and Anthropic. In 2025, Anthropic co-founder Ben Mann testified that he himself had downloaded the massive pirate library LibGen at Anthropic. He did so only a few months after the company was

We built Jikipedia, a new wiki that compiles Jmail data into exhaustive reports on key figures in the Epstein scandal. It lists all recorded visits to Epstein's estates, each person's possible knowledge of Epstein's crimes and laws that they may have violated. Here is

My son is learning Kannada alphabets and struggling a bit. Made a simple progressive web app to practice hearing the sounds. Kashi recommended Sarvam over Google Text To Speech. Any of you native Kannada speakers want to have a look and let me know if it’s accurate for

If you're a researcher in academia using Keras 3 (PhD student, postdoc, professor...) and you want to train on TPUs, you could receive compute awards from Google for your research. Google is running a new academic grant program, separate from the TPU research cloud. You can

For a computer vision problem we're working on with application in the Wellness space, most standard classification models (ConvNeXt, EfficientNet, etc.) were giving amazing results on the data we had : over 97% precision and recall. But we got huge variations in our initial

Nithin Kamath Nice Nithin! A (long) while ago, Raghotham Sripadraj, nischalhp and I took a swing at solving this exact problem. Here's our attempt at it ⤵️ github.com/deep-learning-…

Coding agents allow us to understand and reason about legacy systems - especially their patches. This has been a huge unlock for me in Healthcare s/w systems. Here's one approach that has worked well. Let me explain this with an example. Osirix is a widely used Viewer on Mac

Garry Tan I am betting and bullish on AI-native agencies. Especially in healthcare, where we are building, it is • highly relationship-driven • legacy systems • tribal knowledge. Deployment takes forever- because of customizations. There is going to be a small set of engineers

I get asked this question a lot: what's your favourite episode of The Diary Of A CEO ever? The answer is fairly easy.. episode 101 with Mo Gawdat. It was our most shared episode ever, not because of fame or celebrity, but because what he said was so profound it changed the way I

We've been solving a bunch of CV problems in the manufacturing and wellness domain, conducting tons of experiments with various CLIP models. MobileClip variants consistently outperform with a lower footprint. Great work by Apple 🔥 cc Bargava