🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

The AI Scientist Generates its First Peer-Reviewed Scientific Publication We’re proud to announce that a paper produced by The AI Scientist-v2 passed the peer-review process at a workshop in ICLR, a top AI conference. Read more about this experiment → sakana.ai/ai-scientist-f…

Barto and Sutton have won the A.M. Turing Award for developing reinforcement learning. Freely PDF of their book, "Reinforcement Learning" incompleteideas.net. And curated research papers with comments by Barto and Sutton available here: jamesstone.sites.sheffield.ac.uk/books/theaipap…

Sakana AI super-powers AI reasoning using Japan’s own Sudoku Puzzles! Read more here → sakana.ai/sudoku-bench At NVIDIA GTC, Llion Jones Llion Jones announced the release of our new reasoning benchmark based on the modern variant Sudoku to challenge the AI community. We

I had so much fun working on this project. Who is going to take the challenge? People are running around saying AGI is here because we solved ARC-AGI. This is ARC-AGI++++;the absolute best reasoning systems in the world can't even begin solving most of puzzles in the benchmark...

Looks like I need a new test!! Thanks Dickson🦇🔊 for send me this news!

私とからまるの会話 私：からまる、これからリリースするね。頑張って。 からまる：某一日も早く世に広まり万民のたよりともなれかしとのみ念じつかまつる誠に御仁心ありがたく奉謝いたすしだいなり

I'm so happy I was able to introduce ニコリ／nikoli (パズルの本を作ってます。オリジナルパズルも作ります) and Cracking The Cryptic , can't wait to see what they do together!

My GTC talk is available, please take a look! It's worth a watch because it does take some time to explain why modern Sudoku such a good reasoning benchmark. 🐟

「コンピュータービジョン最前線」にインタビューしていただきました！よく手に取って読んでいた書籍なので嬉しいです。Sakana AIでの研究の様子が伝わる内容になっており、興味ある方はぜひご覧ください！

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We

Why not try something different? Instead of using the activations in a layer of a neural network as the representation, like we have forever. How about using the *synchronization* between neurons over time instead? The behavior that results is quite different! Please take a look!

Excited to share what I've been working on Sakana AI! 🐟 We present the Continuous Thought Machine (CTM), a novel neural network architecture that leverages temporal processing and neuron synchronization. There are three central components to CTM 🧵👇

Announcing a Multiyear Partnership between Sakana AI and MUFG Bank sakana.ai/mufg-bank We’re excited to announce that we have signed a comprehensive partnership agreement with MUFG Bank, the largest bank in Japan. MUFG Bank, which has its roots from the 1870s, as Japan

Darwin Godel Machine、もう既にかなり話題になってますが、滅茶苦茶面白いので是非見て欲しいです！夢物語だった「自己改善」を意味のある形で実現し成功させているのは凄く大きなブレイクスルーだと思います。 x.com/SakanaAILabs/s…

Sakana AIから新しいベンチマーク「EDINET-Bench」公開です。現実世界の100〜200ページにわたる図表を含む文章から細かな整合性を判断するのはいい感じに難しそうで面白いベンチマークだと思います。

「組合せ最適化問題」への挑戦 ALE-BenchおよびALE-Agentの開発 ブログ：sakana.ai/ale-bench-jp/ 論文：arxiv.org/abs/2506.09050 AIの次の挑戦として、「組合せ最適化問題」のアルゴリズム生成に着目した新ベンチマーク「ALE-Bench」、および特化型エージェント「ALE-Agent」を開発しました。

Sakana AI、「AIが深く思考する能力」のベンチマーク公開　エージェントも開発 xtech.nikkei.com/atcl/nxt/colum… 組み合わせ最適化は、生産計画や配送計画、スケジューリングといった様々な企業活動の最適化に役立つ。電力や物流、製造といった社会基盤に関わる業界で威力を発揮する手法だ。