Yash Bhalgat

@ysbhalgat

Gemini Omni @GoogleDeepMind PhD @Oxford_VGG. Past: Senior RS @Qualcomm AI Research, @UMich, @iitbombay All opinions are mine.

United Kingdom

yashbhalgat.github.io

Born 1995

Replies Reposts Media

Pinned
Yash Bhalgat
@ysbhalgat
May 19
Gemini Omni Flash is officially out! 🚀 It has been incredible working on the model pretraining, landing cool features in video editing, and pushing the 3D/camera capabilities in Omni. Seeing it live at #GoogleIO is surreal. 😍😇 Took countless days of no to little sleep. But
00:00
Made with AI
9
Yash Bhalgat
@ysbhalgat
3h
the community notes on this post are epic 🤣🤣🤣 ofcourse, I had to mark them as "helpful" 🫶
Redd
@ReddCinema
Sep 8
00:10
BREAKING NEWS🚨: Iran shoots down an F-15 with a man on a flying carpet 😱😱
Readers added context
It is not F-15. It is F35 globalmilitary.net/compare/aircra…
Yash Bhalgat
@ysbhalgat
4h
I wrote about this an year ago and this is still relevant today: yashbhalgat.github.io/blog/phd-or-no… To December PhD applicants: if your goal is to work on frontiers of video, 3D, or large multimodal models, take a hard look at the landscape before you sign up for a 5-year commitment.
Yash Bhalgat
@ysbhalgat
5h
I am glad @Michael_J_Black acknowledges this. I have myself been contemplating the futility of academic research for a year now. However, I see many academics still in denial, tweaking niche approaches for problems already being solved by better multimodal data and scale.
1
Yash Bhalgat
@ysbhalgat
5h
I am glad @Michael_J_Black acknowledges this. I have myself been contemplating the futility of academic research for a year now. However, I see many academics still in denial, tweaking niche approaches for problems already being solved by better multimodal data and scale.
Michael Black
@Michael_J_Black
15h
Replying to @yuliangxiu
We are presenting VIGA at ECCV this week. It was the first agentic approach to do this. But now I wonder why we are presenting it given how quickly the field has moved. It is making me reevaluate the role of academic vision research. fugtemypt123.github.io/VIGA-website/
1
Yash Bhalgat
@ysbhalgat
Sep 8
stoked about this result. :) this and the saga around this development will change a LOT of things. 🙃
OpenAI
@OpenAI
Sep 8
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem
1