Abheesht Sharma

My toxic trait is that I get too attached to projects.

New function calling guide with Gemma 3 & KerasHub -- really nice to be able to run locally.

Officially validated IMO gold medal, purely via search in token space, achieved in 4.5 hrs (unclear at what compute cost). The solutions read nicely as well deepmind.google/discover/blog/…

From GPT to MoE: I reviewed & compared the main LLMs of 2025 in terms of their architectural design from DeepSeek-V3 to Kimi 2. Multi-head Latent Attention, sliding window attention, new Post- & Pre-Norm placements, NoPE, shared-expert MoEs, and more... magazine.sebastianraschka.com/p/the-big-llm-…

Sometimes the most productive thing you can do when youre stuck on smth is to just walk away and come back later.

Pair programming accelerated software development, could "pair science" do the same for research? What if you had an AI lab partner that never sleeps? 🔬🤖 I built LabMind, an 🧪 AI co-scientist that provides 24/7 monitoring, real-time insights, and powerful simulation…

A problem well-defined is a problem half-solved. The ARC-AGI 3 benchmark itself really stood out in how precisely it frames the challenge to achieve AGI. Now for the other half....📚🔍🧐 @arcprize @fchollet @GregKamradt @mikeknoop #AI #ARCAGI #AGIRevolution

KerasHub now includes HGNetV2! We’re excited to bring the high-efficiency, high-accuracy HGNetV2 image classification backbone into KerasHub’s model family. Model details and quickstart notebook are available on Kaggle: kaggle.com/models/keras/h… #keras #kerashub #HGNet

Someone really shouldn't have let her near the flying Ford Anglia. Ron, this is on you.

Nah, that’s horrible umpiring. Dead in front. Should have been given out on the field.

Interesting, the CEO of @Kimi_Moonshot was the first author of XLNet and TransformerXL Both of which were among the first models added to the @huggingface Transformers library in 2019

When you’re trying to focus at work but your headphones are blasting “Toofanon ko cheer ke, manzilon ko cheen le”, and it’s this heady concoction of KK’s towering vocals, Salim-Sulaiman’s soaring music, and Irfan Siddiqui’s rousing lyrics, and you are all pumped up - but it’s…

Open Models Meetup in Bangalore! 🥳 In just a few weeks, speakers from Google DeepMind's Gemma team, as well as amazing open source contributors, will give some great talks! See you there! Sign up at docs.google.com/forms/d/e/1FAI…

Lucky to get to work on Keras!

Was watching this Netflix documentary on black holes, and learned that Sasha Haco - a theoretical physicist who worked with Stephen Hawking on attempting to resolve the black hole information paradox - is now CEO of an online content moderation startup. Wild pivot.

Keep up the amazing work! 👊

I’ve lost count of how many times I’ve watched Apollo 13 - both the movie and the documentaries. Perhaps even more than Apollo 11. Still leaves me awestruck and inspired every single time.