Good reading on AI alignment, I've been wondering how one could steer LLMs with an equivalent of Three Laws of Robotics https://t.co/82X9F93qRw

— Andrej Karpathy (@karpathy) December 17, 2022

Be sure to catch the end of this clever @NASA "climate spiral" video showing monthly global temperature changes between 1880 and 2021https://t.co/pj6Hucyv7I Sixth in a 🧵 of #2022TopTweets pic.twitter.com/MVs8Z3yZeO

— Conrad Hackett (@conradhackett) December 17, 2022

“It is the more personal pain of seeing a 5% GPU utilization number in production. I am offended by it.” 🔥

—@ID_AA_Carmack, reflecting on his resignation from Facebook. pic.twitter.com/P0D6zODNit

— hardmaru (@hardmaru) December 17, 2022

You can now run Docker with Spaces! 🤯

What does that mean? @elixirlang Phoenix, Dash, FastAPI, Shiny, and your favorite ML tools can now be used with Spaces! 🚀

What will you build? https://t.co/WYGjymGd2F

— Hugging Face (@huggingface) December 16, 2022

NeRF-Art: Text-Driven Neural Radiance Fields Stylization
abs: https://t.co/VVU1EpjrF5
project page: https://t.co/6N4wAxEP4F
github: https://t.co/YyCouPYBdy pic.twitter.com/ErGZgVnPqc

— AK (@_akhaliq) December 16, 2022

MetaPortrait: Identity-Preserving Talking Head Generation with Fast Personalized Adaptation
abs: https://t.co/eEnRIhFH00
project page: https://t.co/TEBWNqM7pR pic.twitter.com/838iPtFKOS

— AK (@_akhaliq) December 16, 2022

Rethinking Vision Transformers for MobileNet Size and Speed
abs: https://t.co/yEEjhowiau
github: https://t.co/2k9fYH8QAJ pic.twitter.com/3PJMJ6PNu5

— AK (@_akhaliq) December 16, 2022

FlexiViT: One Model for All Patch Sizes
abs: https://t.co/YEuJBKgJar
github: https://t.co/LrAejNdeG8 pic.twitter.com/ogUFBtmMg7

— AK (@_akhaliq) December 16, 2022

"We present gen. Parameter-Efficient Finetuning framework for tuning LLM with only 𝟬.𝟭%-𝟬.𝟮% of parameters using mix of adaptation modules -> achieve new 𝗦𝗢𝗧𝗔 > standard tuning on both NLU & NLG tasks.
Paper: https://t.co/yE5YeSBu8m
Code&Models: https://t.co/rT6Q0vES45

— Leo Boytsov (@srchvrs) December 16, 2022

Image-and-Language Understanding from Pixels Only
abs: https://t.co/E9fOot76FZ pic.twitter.com/H3x6GMMXqh

— AK (@_akhaliq) December 16, 2022

I don't claim TikTok is "better". I'm just trying to explain its success. There's a dark side to a scrolling UX that relies on our automatic actions rather than our explicit choices: it feeds our basest impulses. I've previously written about it here: https://t.co/wI5ezce9JT

— Arvind Narayanan @randomwalker@mastodon.social (@random_walker) December 15, 2022

Imagen Editor and EditBench: Advancing and Evaluating Text-Guided Image Inpainting
abs: https://t.co/WDKOzrdC6m
project page: https://t.co/qsnnbjrlhz pic.twitter.com/qWKJJMJ3pV

— AK (@_akhaliq) December 15, 2022