Rogério Guimarães (@rogerioagjr) 's Twitter Profile
Rogério Guimarães

@rogerioagjr

PhD Student in Computation & Neural Systems @Caltech. Studied CS, Linguistics & Philosophy @MIT. Excited to explore how all these fields can work together!

ID: 4605684514

linkhttp://rogeriojr.com calendar_today26-12-2015 01:12:01

248 Tweet

397 Followers

495 Following

Mostafa Dehghani (@m__dehghani) 's Twitter Profile Photo

What do you think are the primary limitations or design choices that feel unnatural when it comes to using Transformers for computer vision (images, videos, ...)?

What do you think are the primary limitations or design choices that feel unnatural when it comes to using Transformers for computer vision (images, videos, ...)?
Rogério Guimarães (@rogerioagjr) 's Twitter Profile Photo

I started running in January as a new year resolution and this weekend I finished The SF Marathon! Hope it’s the first of many, as my time has lots of room for improvement 😅

I started running in January as a new year resolution and this weekend I finished <a href="/theSFmarathon/">The SF Marathon</a>! 

Hope it’s the first of many, as my time has lots of room for improvement 😅
Suzanne Stathatos (@suzstathatos) 's Twitter Profile Photo

TA'ing this class 🤖 this past spring was a lot of work, but very fun! Very cool to see it featured in Caltech Magazine. 📚 Very grateful to have had such an incredible teaching team Georgia Gkioxari Pietro Perona Laure Delisle Neehar Kondapaneni Zihui (Ray) Wu Rogério Guimarães

Pablo Samuel Castro (@pcastr) 's Twitter Profile Photo

You can read a paper with extreme detail, or watch a video of someone explaining an algorithm or method, that you might feel you "get it". But you don't really fully "get it" until you implement it and run it yourself. Only then do you really achieve enlightened expertise.

Rogério Guimarães (@rogerioagjr) 's Twitter Profile Photo

We’re thrilled to announce our work has been accepted to #CVPR2024! We analyze the role of text prompts when repurposing diffusion models for different vision tasks and achieve SOTA results, especially in cross-domain tasks, with automated captioning. See you all in Seattle!

Rogério Guimarães (@rogerioagjr) 's Twitter Profile Photo

We've just released the code for TADP, our #CVPR2024 paper! It’s ready for benchmarking on popular tasks if you need baselines for #NeurIPS2024. We also released a Colab demo showing cross-domain inference in WaterColor with a model trained in Pascal. Links in our project page!

We've just released the code for TADP, our #CVPR2024 paper! It’s ready for benchmarking on popular tasks if you need baselines for #NeurIPS2024.

We also released a Colab demo showing cross-domain inference in WaterColor with a model trained in Pascal.

Links in our project page!
Neehar Kondapaneni (@therealpaneni) 's Twitter Profile Photo

The final version of our ICLR 2024 Re-Align workshop paper is out -- along with code and our new dataset! Do models and humans make use of the same visual concepts to arrive at a decision? vision.caltech.edu/dcne/

The final version of our ICLR 2024 Re-Align workshop paper is out -- along with code and our new dataset! Do models and humans make use of the same visual concepts to arrive at a decision? vision.caltech.edu/dcne/
Rogério Guimarães (@rogerioagjr) 's Twitter Profile Photo

Humans sure are smart, but animals with much smaller brains continue to surprise us with their learning skills! Come see our work in #CCN2024 showing that mice can master many complex mazes in a single day and remember them, even those without a cortex! 2024.ccneuro.org/poster/?id=485

Ueli Rutishauser (@uelirutishauser) 's Twitter Profile Photo

Delighted our latest finding! We discovered that abstract representations emerge in the human hippocampus when learning to perform inference. This change in neural geometry is due to disentanglement of discovered latent and observable variables. nature nature.com/articles/s4158…