📢 #adversarial #data #challenge for text-to-image #GenerativeAI /w @kaggle @MLCommons & #DataPerf 👀 https://t.co/i5CV0ctLmr &✍️ a paper with your📊 to the Art of Safety Workshop (virtual) 📌 https://t.co/0YoyIFb42J at @aaclmeeting 🗓️Deadline 25 Aug 2023

You went 🍌🍌 for Nano Banana. Now, meet Nano Banana Pro.  It’s SOTA for image generation + editing with more advanced world knowledge, text rendering, precision + controls. Built on Gemini 3, it’s really good at complex infographics - much like how engineers see the world:)

We processed over 1.3 Quadrillion tokens last month - that's 1,300,000,000,000,000 tokens! or to put it another way that's 500M tokens a second or 1.8 Trillion tokens an hour... 🤯

Preparing a dataset for @NeurIPSConf? Check out Croissant-integrated repositories like Kaggle Datasets to streamline metadata generation. Let’s set a new standard for dataset contributions!

I'm excited about the updates to dataset submission requirements and best practices we introduced this year!!!

Amazing!! Congratulations @cawelty !!

feeling pretty excited about being part of Expanding AI Overviews and introducing AI Mode @google

📢 Join Google DeepMind's DEER team & shape the future of #ResponsibleAI! We're hiring a #ResearchScientist (Fixed Term Contract, 12 month) to tackle fairness in multi-modal AI. Make a real-world impact! Apply by Feb 7th: https://t.co/RKSz8AUa0V @AliciaVParrish

Working with @laroyo and Lingjuan Lyu to make this talk and this year’s track happen has been wonderfully rewarding!

@jmtomczak @NeurIPSConf @DaniCMBelg @ulrichpaquet @laroyo @FrancescoLocat8 Congrats! You've all done an amazing job organizing (the yet again largest) NeurIPS this year 👏 Was great to peek a bit behind the curtain at the VIP reception today 😁 What a momentous effort!

Achieving your goals is impossible without being part of a great team! It feels amazing to be in front of thousands of researchers at @NeurIPSConf, knowing there are outstanding people who have your back.

Congratulations 🎉 @mattlease

Ted Chiang's latest on AI. I dont entirely agree or disagree, but he weaves a good narrative all the same. @ylecun for visibility.

We won OUTSTANDING PAPER at #ACL2024 for our work on evaluating values and opinions in LLMs 🥳 Thank you to the reviewers and awards committee, and again to my amazing co-authors, especially joint first author @vjhofmann Check out the poster + paper + full author list below 👇

This is ongoing work, and please do reach out if you have thoughts, comments, criticisms about this, or any of these papers, or thoughts on potential future collaborations.

In prior work, we studied ways to disentangle disagreements in annotations along demographic lines and integrate them into ML data/model pipelines. Here, we focus on how to reliably measure distinct group-level perspectives that differ significantly from those of other groups.

More specifically, we propose GRASP, a comprehensive and extensible disagreement analysis framework to measure group association in perspectives --- considering both in-group and cross-group cohesion --- among different rater sub-groups.

This is one of the many papers we have published over the last year, coming out of an amazing 2-3 year long (and ongoing) collaboration with an awesome set of collaborators within and beyond Google!

I'll be at #naacl2024 (@naaclmeeting) this week; come say hi if you are too! I'll be presenting this paper on measuring (sub-)group associations in annotation disagreements; joint work w/ Chris Homan @laroyo @aidaa @AliciaVParrish @alxndrt @blahtino @justaturbo & @gregserapio

Is a picture of a bird a bird? A mixed-methods approach to understanding diverse human perspectives and ambiguity in machine vision models https://t.co/gwwqTWscqX By @AliciaVParrish, Susan Hao, Sarah Laszlo, @laroyo (2/17)