Dalin Guo

currently @GoogleDeepMind previously @twitter

[Google DeepMind @GoogleDeepMind]

We’re reimagining a 50-year-old interface - the mouse pointer - with AI. 🖱️

These experimental demos show how people can intuitively direct Gemini on their screens using motion, speech, and natural shorthand to get things done 🧵

[Google @Google]

Coming soon, we’ll bring new multi-step reasoning capabilities to Google Search. It breaks your bigger question down into parts and figures out which problems to solve and in what order, so research that might've taken you minutes or even hours can be done in seconds. #GoogleIO

[bioRxiv Neuroscience @biorxiv_neursci]

Multiple and subject-specific roles of uncertainty in reward-guided decision-making https://t.co/S8StypsIs7 #biorxiv_neursci

[Google DeepMind @GoogleDeepMind]

Introducing SIMA: the first generalist AI agent to follow natural-language instructions in a broad range of 3D virtual environments and video games. 🕹️

It can complete tasks similar to a human, and outperforms an agent trained in just one setting. 🧵 https://t.co/qz3IxzUpto

[Google DeepMind @GoogleDeepMind]

Introducing FunSearch in @Nature: a method using large language models to search for new solutions in mathematics & computer science. 🔍

It pairs the creativity of an LLM with an automated evaluator to guard against hallucinations and incorrect ideas. 🧵 https://t.co/MC5ttgvZeM

[Google DeepMind @GoogleDeepMind]

We’re excited to announce 𝗚𝗲𝗺𝗶𝗻𝗶: @Google’s largest and most capable AI model.

Built to be natively multimodal, it can understand and operate across text, code, audio, image and video - and achieves state-of-the-art performance across many tasks. 🧵 https://t.co/mwHZTDTBuG

[Kim Stachenfeld (neurokim.bsky.social) @neuro_kim]

Applications open for student researchers (formerly interns)!
https://t.co/oHSaOKb39u

There will likely be projects with me+colleagues (including @kevinjmiller10!) on comp neuro, and ML directions on memory+retrieval. Locations in NYC/London.

Reach out if any questions!

[Demis Hassabis @demishassabis]

The Gemini era is here. Thrilled to launch Gemini 1.0, our most capable & general AI model. Built to be natively multimodal, it can understand many types of info. Efficient & flexible, it comes in 3 sizes each best-in-class & optimized for different uses https://t.co/VUu1277bC2

[Demis Hassabis @demishassabis]

Thrilled to share #Lyria, the world's most sophisticated AI music generation system. From just a text prompt Lyria produces compelling music & vocals. Also: building new Music AI tools for artists to amplify creativity in partnership w/YT & music industry https://t.co/CMttmLPjoC

[Christoph Feichtenhofer @cfeichtenhofer]

At #ICML2023, Chay (@wrong_whp) will present Hiera, a hierarchical vision transformer that is fast, powerful, and simple. Code+models at: https://t.co/barBBsup7N

If interested, please come to the oral presentation on Tue 25 Jul 5:30pm HST or poster #219 on Wed 26 Jul 2pm HST.

[Google DeepMind @GoogleDeepMind]

As society becomes more digital, it’s critical to improve the code powering the world's computing.

Today in @Nature, we present AlphaDev, an AI system using reinforcement learning to discover enhanced computer science algorithms.

How does it work? 🧵 https://t.co/2ukGaPey5w

[AI at Meta @AIatMeta]

📝 New research from Meta AI — Hiera is an extremely simple hierarchical vision transformer that's both more accurate than previous models + significantly faster at inference and during training.

Paper ➡️ https://t.co/mPN3nPra6g

[AK @_akhaliq]

Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles

abs: https://t.co/HdZSiYNUcb

Modern hierarchical vision transformers have added several vision-specific components in the pursuit of supervised classification performance. While these components lead to effective accuracies and attractive FLOP counts, the added complexity actually makes these transformers slower than their vanilla ViT counterparts. In this paper, we argue that this additional bulk is unnecessary. By pretraining with a strong visual pretext task (MAE), we can strip out all the bells-and-whistles from a state-of-the-art multi-stage vision transformer without losing accuracy. In the process, we create Hiera, an extremely simple hierarchical vision transformer that is more accurate than previous models while being significantly faster both at inference and during training. We evaluate Hiera on a variety of tasks for image and video recognition

[CosyneMeeting @CosyneMeeting]

Dear Cosyne Community,

Last week, An Wu, a postdoc in the Komiyama lab at UCSD
who presented at this year’s meeting, has been reported missing. Our current information is that after Cosyne, she stayed at a building that burnt down in fire in Old Montreal.

#cosyne2023 1/2

[Corey Zhou @cor2yz]

Thrilled to share this new preprint w/ @supergrrl007 @nathanieldaw @marcelomattar: https://t.co/eX9Yr7bpWO
Starting with the question “what is episodic memory for?”, we propose an algorithmic theory of decision making where model-based evaluation is achieved by episodic recall.

[Arya Mazumdar @MountainOfMoon]

The @HDSIUCSD and ECE dept. at @UCSanDiego have a joint faculty position in the area of Machine Learning Theory/System focusing on Optimization methods. Please apply?
https://t.co/An6zW1aYJT

[Dalin Guo @dalinguo]

@alykhantejani I’m binge-watching TV so far 😂

[Dalin Guo @dalinguo]

#lovewhereyouworked 🫡

[Wenzhe Shi 🐕🐎 @trustswz]

We describe a method for maximising long term engagements by using model-based reinforcement learning. The policy is then used to make decisions about whether to send push notifications or not. More details are in the paper if you are interested. https://t.co/Ww6lKHoim9

[Rosanne Liu @savvyRL]

In 2020, my life was turned upside down -- I was encountering failure after failure and in deep depression. But that also turned out to be the most enlightening period. In October I was invited to give a talk at Google (thanks to @orf_bnw). It's out today: https://t.co/hYFLnH7JHV