ARC Prize is evolving into a full-fledged non-profit foundation, to further our mission of guiding and accelerating research progress towards AGI. A special thanks to Greg Kamradt , who will be leading the charge as ARC Prize President. His dedication and expertise, honed during

where the world stands on the paris climate agreement, as of today:

Is this the first time one can buy something other than GPUs to predict protein structures fast? Computed on Tenstorrent Wormhole in 1 minute and 25 seconds.

Why is the ratio of S&P500/VOO ~ 10.9 instead of a nice round 10, like for SPY?

Nobody told me that in May of 2024, Gaitsgory, D., Raskin, S., et al. proved the Geometric Langlands correspondence. people.mpim-bonn.mpg.de/gaitsgde/GLC/

I'm getting more confident that Tenstorrent boards can become open & better alternatives to GPUs in biotech. Now that I parallelized triangle self-attention of AlphaFold 3 across both Tenstorrent Wormhole processors, we can predict the structure of proteins with up to ~750 amino

Much of the field obsesses over end-to-end learning. But strong generalization requires compositionality: building modular, reusable abstractions, and reassembling them on the fly when faced with novelty. The models of the future won't be just pipes, they will be Lego castles.

The Tenstorrent business model. Some stuff is given away. Cloud is free for demos/PoC. Sell IP. Sell chips. Sell systems.

Fascinating talks by Jim Keller and others this morning at Tenstorrent Dev Day, taking about the benefits and importance of open #RISCV, with the intention that their new technologies will be some of the most powerful RISC-V International based CPUs with a focus on openness

hoping for an open future openfuture.tenstorrent.com Tenstorrent

I called by Tenstorrent for a chat with CEO Jim Keller Jim Keller about his commitment to open-source, the training systems he's building, and why he wants to do the opposite of whatever Jensen does: eetimes.com/jim-keller-wha…

I have bittersweet news to share. Yesterday we merged a PR deprecating TensorFlow and Flax support in transformers. Going forward, we're focusing all our efforts on PyTorch to remove a lot of the bloating in the transformers library. Expect a simpler toolkit, across the board.

The BBB is going to ruin poker in the US. Please note that the relevant part was pushed by Mike Crapo, Idaho Senator. frontofficesports.com/mike-crapo-gam…

openai.com/open-models

As a guest on the Creatives Podcast, I discuss my transition from physics to AI, and what it means to be creative in a world with AI. youtube.com/watch?v=r2gVUG…

Amazing..

The GPT 5 launch included a chart showing 52.8 as a bigger number than 69.1, which in turn is shown as the same magnitude as 30.8. Not quite ASI…

The 3rd edition of my book Deep Learning with Python is being printed right now, and will be in bookstores within 2 weeks. You can order it now from Amazon or from Manning. This time, we're also releasing the whole thing as a 100% free website. I don't care if it reduces book

i think at least right now for ml researchers there's an art to using ai, that probably will change like monthly. if you dont use it entirely you're going unnecessarily slow but if you use it wrong your whole research direction will probably be built on incorrect slop

The term “AGI” is currently a vague, moving goalpost. To ground the discussion, we propose a comprehensive, testable definition of AGI. Using it, we can quantify progress: GPT-4 (2023) was 27% of the way to AGI. GPT-5 (2025) is 58%. Here’s how we define and measure it: 🧵