Very excited to share the outcome of my internship @Google where we propose a new elastic visual tokenization architecture:
PARCEL: Pool-Anchored Resampling with Conditioned Elastic Queries for Efficient Vision-Language Understanding
https://t.co/9d2XuqiKpB
Thread below π§΅π
@YouJiacheng Currently the tax rate is lower than the best risk free interest rate you can get on CHF. It also depends on where you live so you can optimize your residency based on your net worth.
@NandoDF@WenhuChen Would be cool to have a Gemma equivalent of releases from MS. Open source needs to live on to educate the next generation of builders.
@chrisoffner3d Itβs not the most crazy example. Some banks used to physically mail you when a new device logged into your account π. That information could totally wait the two weeks it took to receive the letter.
Not Americans but international teams developed convnets, alexnet, attention, AlphaGo, neural LMs, AlphaCode, AlphaFold, transformers, RL, etc, etc.
This war mongering CEO does not represent the Americans who helped develop AI either. They have greater ideals.
It is sad that my colleagues and I developed the science and tech being used by these money and power hungry despicable people.
We need a moratorium on AI weapons. And international institutions that can enforce it.
@xidulu SR process is mostly team led, if you reach out to teams you are interested in working with, you will have a better luck
Most teams already have a list of candidate they want
Unlike the intern program which is more central, SR is decentralised
@Neiluss_@chrisoffner3d TUM hired some great young faculty in the last few years particularly on deep learning
ETH went the more specialised per topic faculty
Research wise you canβt go wrong with either. My teaching experience is too old now to comment on it. At ETH I only did my PhD.
@chrisoffner3d At the same time they lacked depth, I relied on open sources courses from Stanford to self study
Our deep learning course was half the depth of Stanford 231N
@chrisoffner3d My professors at TUM used to actively shit on deep learning back in 2018 saying it is unlikely to lead to meaningful breakthroughs because itβs not interpretable, that aged well π
But to be fair we had some great deep learning courses coming from newer faculty at the same time
Excited to share what we have been up to in image text embedding models. SigLIP 2 is the most powerful encoder for most open vocabulary computer vision and MMLLM tasks. Checkpoints are open sourced and we look forward to what the community achieves with these.
Joint Statement: Apple and Google have entered into a multi-year collaboration under which the next generation of Apple Foundation Models will be based on Google's Gemini models and cloud technology. These models will help power future Apple Intelligence features, including a more personalized Siri coming this year.
After careful evaluation, Apple determined that Google's Al technology provides the most capable foundation for Apple Foundation Models and is excited about the innovative new experiences it will unlock for Apple users. Apple Intelligence will continue to run on Apple devices and Private Cloud Compute, while maintaining Apple's industry-leading privacy standards.