Priyank Jaini
@priyankjaini
Research Scientist @GoogleDeepMind | Prev: Postdoc @AmlabUva, PhD @UWCheritonCS, Undergrad @IITKanpur | opinions my own |
Do generative video models learn physical principles from watching videos? Very excited to introduce the Physics-IQ benchmark, a challenging dataset of real-world videos designed to test physical understanding of video models. Webpage: physics-iq.github.io
Super excited to finally share what I had been working on along with an amazing team of collaborators these past few months. Particularly on computer control, multimodal memory, and using memory systems for guidance. #GoogleIO #ProjectAstra
Last year, we introduced Project Astra: a research prototype exploring capabilities for a universal AI assistant. 🤝 We’ve been making it even better with improved voice output, memory and computer control - so it can be more personalized and proactive. Take a look ↓ #GoogleIO
🚨 New Preprint! 🚨 We explore Amortized In-Context Bayesian Posterior Estimation with Niels, @g_lajoie_, @priyankjaini & @marcusabrubaker! 🔥 Amortized Conditional Modeling = key to success in large-scale models! We use it to estimate posteriors 🔑 📄 arxiv.org/abs/2502.06601
Will be presenting this work as a spotlight at #ICLR on Thursday. Please drop by our poster or DM me to know more about intriguing properties of generative classifiers. Joint work with @clark_kev and Robert Geirhos.
Excited to share our work on studying perceptual properties of #StableDiffusion, #Imagen & #Parti by evaluating them as zero-shot classifiers! We report four intriguing emergent properties of these models. 🧵👇 w. Robert Geirhos* & @clark_kev Paper: arxiv.org/abs/2309.16779
I will be presenting this work at the poster session tomorrow at #NeurIPS23 at 10:45am at poster #1913. Please drop by if you are interested to learn more about zero-shot capabilities of text-to-image diffusion models. w. @clark_kev
.@clark_kev & I are excited to share our new work on studying Imagen by evaluating it as a zero-shot classifier! Highlights include Imagen achieving SoTA on Stylized Imagenet and being able to perform attribute binding in certain settings unlike CLIP arxiv.org/abs/2303.15233 🧵👇
I’ll be at #NeurIPS to present this work on zero-shot classification capabilities of text-to-image diffusion models. Would love to meet people interested in multimodal models, probabilistic generative models and application in vision.
.@clark_kev & I are excited to share our new work on studying Imagen by evaluating it as a zero-shot classifier! Highlights include Imagen achieving SoTA on Stylized Imagenet and being able to perform attribute binding in certain settings unlike CLIP arxiv.org/abs/2303.15233 🧵👇
We have a student researcher opportunity in our team @GoogleDeepMind in Toronto 🍁 If you’re excited about research on diffusion models, and generative video models, please fill the form : forms.gle/auNq61N35AvTZS… and apply here: deepmind.google/about/careers/…
