🔥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

"To be clear, no one has proposed using ROCK inhibitors to treat dry AMD in the literature before, as far as we can find..." It's so unprecedented, there's even a review article... sciencedirect.com/science/articl…

I'm pretty amazed by everything people at GDM have been building ngl 🤯 (I should keep up more internally but IO was a great way to see what's new)

So who signed off on the acquisition?

Tfw you are close to AGI but markdown rendering is still somehow broken 😅

Spaguettis are so cooked. But flamenco is so back! "A dog dressed as a female flamenco dancer dancing flamenco on a tablao in a bar in Seville." 😅😂

DumPy: Like NumPy except it's OK if you're dum

Introducing the next generation: Claude Opus 4 and Claude Sonnet 4. Claude Opus 4 is our most powerful model yet, and the world’s best coding model. Claude Sonnet 4 is a significant upgrade from its predecessor, delivering superior coding and reasoning.

opus 4 is here—“o4”, if you will

Prompt Theory (Made with Veo 3) What if AI-generated characters refused to believe they were AI-generated?

1. We often observe power laws between loss and compute: loss = a * flops ^ b + c 2. Models are rapidly becoming more efficient, i.e. use less compute to reach the same loss But: which innovations actually change the exponent in the power law (b) vs change only the constant (a)?

arxiv.org/abs/2505.16932 everyone told me even SVD won't improve loss so I only spent a bit effort on improving the iters. After reading leloy! 's post last year, I had the intuition that greedy "contraction" will be a good solution, but didn't know it's optimal.

I read through these slides and felt like I was transported back to 2018. Having been in this spot years ago, thinking about what John & team are thinking about, I can't help but feel like they will learn the same lesson I did the hard way.

The French Ministry of Culture released 175k high-quality arena-style preferences It's exactly the type of data LMSYS stopped releasing. They created their own chatbot arena with 55 models and open-sourced everything. Incredible work! 🤗 Dataset: huggingface.co/datasets/minis…

THE WAY OF CODE, a project by @rickrubin in collaboration with Anthropic:

Strong recommend for this book and the JAX/TPU docs, even if you are using Torch / GPUs. Clean notation and mental model for some challenging ideas. github.com/jax-ml/scaling… github.com/jax-ml/scaling… docs.jax.dev/en/latest/note…

Hearing a lot of supposedly smart people complain about Anthropic averaging pass@1 over 10 trials and "giving themselves an advantage", but has no one heard of regression to the mean?

few realize this: Anthropic's coding focus isn’t a step back from AGI it's the quickest path to it

Steven Pinker injects some facts and much-needed sanity in the debates around Harvard and American academia. nytimes.com/2025/05/23/opi…