Python 3.13 is crazy. Multiple interpreters(PEP 734), with per-interpreter GIL (PEP 684), optional GIL (PEP 703) and JIT compilation (PEP 744). Beta 4 comes out tomorrow and RC1 on 30 July.

Gemini #1 now on LMSys !

Lifehack: Scrape the (title, abstract) from the reviewer bidding console in open review (I used selenium) and put them in NotebookLM. Much more informed bidding than adhoc keyword searches.

Yeah. Not good. sciencedirect.com/science/articl…

RIP Felix docs.google.com/document/d/1aE…

As the blessed month of Ramzan begins, may it bring peace and harmony in our society. This sacred month epitomises reflection, gratitude and devotion, also reminding us of the values of compassion, kindness and service. Ramzan Mubarak!

I still support Ukraine personally. All the arguments I've seen for betraying Ukraine ($, peace, nuclear war, stalemate) are bogus after looking into the details.

I got hit by some rather sudden and extreme financial hardship so if anyone is in need of remote wetlab contract research, strictly BSL1, do let me know. Currently scrambling for gigs. Plant, Bacterial, Archaeal Bioeng Custom Lab Hardware Turn Key Genetic Design Please RT 💚

I have been using Alexa+ in my home across 5 different devices for months and am very impressed at the capabilities. Whether it’s 10x or 100X better doesn’t matter, but it’s just that much better than its predecessor. It remembered my food choices, my exercise routines, my

Pretty on point youtu.be/IXR9PByA9SY?si…

Transformers are still dominating the LLM scene but we show that higher throughput alternatives exist which are just as strong! Grateful to have a part in Nemotron-H Reasoning effort. 🙏 Technical report will be out soon, stay tuned!

Yesterday, our AI communication app helped a nonverbal autistic child talk about ear pain she'd been experiencing. She had been crying and upset, but no one knew why. She was diagnosed with an inner ear infection and is now getting treatment. This is why we build🧡

6) Last, we introduce "Concept Hacking" to reveal core knowledge deficiencies in the control experiment set-up. Concept Hacking systematically manipulates the task-relevant features while preserving all task-irrelevant conditions ... (10/n)

Has anyone tried running AI models (CNNs/LLMs, ViTs/ Diffusion) on weird chips? Edge: Qualcomm AR1, Ambarella, TensTorrent Cloud: Trainium, Inferentia, AMD Or even just porting Ampere → Hopper → Blackwell? Curious: how painful was it? Did it kill your project before it started?

🚨 Announcing Generalized Correctness Models (GCMs) 🚨Finding that LLMs have little self knowledge about their own correctness, we train an 8B GCM to predict correctness of many models, which is more accurate than training model-specific CMs, and outperforms a larger

🧐 LLMs aren’t great at judging their own correctness. ❗But history across models helps! We present Generalized Correctness Models (GCMs), which learn to predict correctness based on history, outperforming model-specific correctness and larger models' self-confidence.

Deciding between extending Art, SkyRL, and OpenRLHF for multi-objective RL . Needed process supervision + LoRA so Verl, trl, nemorl etc. are out. Anyone shipped with these? Which has the least footguns, good acceptance of PRs? Kyle Corbitt Sumanth Hegde Ankit Maloo #skyrl #openrlhf #roll