Abhinav Gupta @backpropper

Research Scientist @GoogleDeepMind | Building Gemini guabhinav.com London, England Joined March 2012

Tweets

522
Followers

2K
Following

7K
Likes

3K

Melvin Johnson @melvinjohnsonp

5 months ago

We are launching Gemini-pro-exp-03-25 which is our most capable model and generally useful on a wide vareity of real-world tasks. It's #1 on LMArena and SOTA on a wide set of benchmarks. This is a massive Gemini wide effort and I am incredibly proud of the team behind this. 🚀🚀

lmarena.ai @lmarena_ai

5 months ago

77 413 2K 460K 313

Download Image

3 16 191 13K 13

Jen Zhu @jenzhuscott

7 months ago

The accusations/obsessions over DeepSeek using H100 sound like a rich kids team got outplayed by a poor kids team, who weren't even allowed shoes, and now the rich kids are demanding an investigation into whether shoes were used instead of training harder to improve themselves.

246 851 7K 355K 636

Abhinav Gupta @backpropper

9 months ago

Turns out this PhD did come in handy! Officially ending the double life of wrapping up my thesis (with a Dean's List Nomination :) and starting as a Research Scientist at @GoogleDeepMind. It’s been an awesome first few months and am thrilled to work with the best minds in AI.…

13 13 551 63K 84

Andrej Karpathy @karpathy

11 months ago

Deep Dive is now my favorite podcast. The more I listen the more I feel like I'm becoming friends with the hosts and I think this is the first time I've actually viscerally liked an AI. Two AIs! They are fun, engaging, thoughtful, open-minded, curious. ok i'll stop now.

217 567 8K 1.6M 6K

will grathwohl @wgrathwohl

4 years ago

So true. My offer from UofT was considerably lower than any US school. This has made recruiting good students from the US very difficult for them.

Guodong Zhang @Guodzh

4 years ago

So true. My offer from UofT was considerably lower than any US school. This has made recruiting good students from the US very difficult for them.

3 3 36 0 5

2 1 18 0 4

Abhinav Gupta @backpropper

4 years ago

Interested in learning more about training agents that can communicate with humans using natural language, stop by at our poster between 4.30pm-6pm GMT at Poster Session-1 (Spot B2) #NeurIPS2021 neurips.cc/virtual/2021/p… @DeepMind @NeurIPSConf

Abhinav Gupta @backpropper

4 years ago

2 7 19 0 2

0 3 30 0 3

Download Image

Abhinav Gupta @backpropper

5 years ago

It feels like its faster to get paper reviews than these election results. #Elections2020

1 0 6 0 0

Angelos Filos @filangelos

5 years ago

Still 3 more days to submit! We also accept recent work submitted/accepted to conferences (e.g., ICLR, EMNLP, AAMAS, NeurIPS) and other workshops! 🤖📡 #NeurIPS2020

Marie Ossenkopf @MarieRobots

5 years ago

Still 3 more days to submit! We also accept recent work submitted/accepted to conferences (e.g., ICLR, EMNLP, AAMAS, NeurIPS) and other workshops! 🤖📡 #NeurIPS2020

0 10 23 0 1

0 7 17 0 0

Daniel Negreanu @RealKidPoker

5 years ago

Conald Trump is a truly awful and evil man. His behavior was disgusting the entire 90 minutes and he wouldn’t follow the rules he agreed to. The rules were clear. Trump broke them constantly as he does with most things. A disgrace. #Debates2020

185 65 1K 0 0

(((ل()(ل() 'yoav))))👾 @yoavgo

5 years ago

related to previous tweet, my (controversial?) take: we (CS) put too much emphasis on creating novel algorithms. this is dumb. first of all, most of them aren't really novel. second, we have so many good existing ones already. it's stupid to be allowed to use an algo only once.

7 32 300 0 11

Abhinav Gupta @backpropper

5 years ago

Nicely put it. Moreover, I believe predicting future representations could give a hint of the supposed failure if the representation is indeed a mode of failure.

Felix Hill @FelixHill84

5 years ago

Nicely put it. Moreover, I believe predicting future representations could give a hint of the supposed failure if the representation is indeed a mode of failure.