🔥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

iOS 26 looks amazing

New iOS design just dropped

Anyone else think that this looks familiar? 🤔 Windows Vista was ceased.

Apparently everyone else already thought so. I was late to the party 😅

Very excited to release our first reasoning model, Magistral. We released the weights of Magistral Small alongside a paper that presents our approach, online RL infrastructure, and findings.

The history of AI from 2012 to present shows each paradigm solving previous limitations while revealing new ones. The next state-of-the-art will be Supervised RL for reasoning. It is fundamentally bottlenecked by the need for verifiable environments. 2/7

Looks interesting! Here is what I did for fast adaptation 5 years ago: arxiv.org/pdf/2009.01803. Curios to see the advancement since then.

The Inception paper arxiv.org/abs/1409.4842 was awarded the Longuet-Higgins prize (Test of time). The architecture represented a significant step forward in inference efficiency especially on CPU and variants of Inception networks were used in Google products for years.

Day 1/5 of #MiniMaxWeek: We’re open-sourcing MiniMax-M1, our latest LLM — setting new standards in long-context reasoning. - World’s longest context window: 1M-token input, 80k-token output - State-of-the-art agentic use among open-source models - RL at unmatched efficiency:

Eighty U.S. Senators are backing tariffs on Russian oil buyers. As MONGOLIA, democracy’s the lone outpost, we face a hard truth: landlocked and bordered by two giants. We have no alternative but to import fuel from the north. I hope our unique position earns an understanding and

𝑨𝒐𝑬 DeepSeek-R1T-Chimera, a 671B hybrid model built by merging only routed experts from DeepSeek-R1 into V3-0324 - No fine-tuning, no distillation - Matches R1 reasoning while using ~40% fewer output tokens (~2.5× more concise) - Fully functional out-of-the-box Method -

Гүүглээс Gemini 2.5 хиймэл оюуны загвар хөгжүүлэлтийн тухай техникал рефорт гарсан шүү: storage.googleapis.com/deepmind-media…

Ah what a great idea! Put teacher at the center of training.

Reasoning, Attention & Memory Workshop @ COLM Submission Deadline: June 23, 2025 -- Today!

Excited to share that our paper on model merging at scale has been accepted to Transactions on Machine Learning Research (TMLR). Huge congrats to my intern Prateek Yadav and our awesome co-authors Jonathan Lai, Alexandra Chronopoulou, Manaal Faruqui, Mohit Bansal, and Tsendsuren 🎉!!

This work got accepted at Transactions on Machine Learning Research (TMLR). Congratulations to Prateek Yadav and my co-authors. Also, thank you to the reviewers and editors for their time.

За ус бдаггүй гадаа бороонд нүцгэн зогсдог юм уу 😅

🎉 We’re excited to introduce BLAB: Brutally Long Audio Bench, the first benchmark for evaluating long-form reasoning in audio LMs across 8 challenging tasks, using 833+ hours of Creative Commons audio. (avg length: 51 minutes).

Two of my former interns Trapit Bansal and Prateek Yadav are at Meta Superintelligence. Excited for what they are gonna build together 🥳

Tsendsuren Prateek Yadav I had tried implementing one of your papers: "Infini Transformer" from scratch: github.com/ArionDas/Infin… But you never shared the code with me when I requested 😄🙌