Introducing a novel mixture-of-experts routing algorithm, called Expert Choice, that can achieve optimal load balancing between experts while allowing heterogeneous token-to-expert mapping. Learn how it’s done at https://t.co/EDomelBL53 pic.twitter.com/Oyaa0wEMlX
— Google AI (@GoogleAI) November 16, 2022
Large Language Models Struggle to Learn Long-Tail Knowledge
— AK (@_akhaliq) November 16, 2022
abs: https://t.co/8WKY88NkfM pic.twitter.com/rulgHJDvyH
Versatile Diffusion: Text, Images and Variations All in One Diffusion Model
— AK (@_akhaliq) November 16, 2022
abs: https://t.co/UjdtkxFY2s
github: https://t.co/N3bQbpI4SB pic.twitter.com/KW8Ix4O91g
“Defective Altruism” by @NathanJRobinson
— hardmaru (@hardmaru) November 15, 2022
I’m far from a leftist-socialist, but this article is a great read: https://t.co/FJvtFayru9
Many are lured into Effective Altruism by caring about the future only to discover the truth that EA people are very preoccupied with AGI x-risk. pic.twitter.com/qx8ZxsGDBB
How gerrymandering can change election outcomes, in one charthttps://t.co/BQ5PpS4c2c pic.twitter.com/jkPVrPVsnh
— Conrad Hackett (@conradhackett) November 15, 2022
Seeing Beyond the Brain: Conditional Diffusion Model with Sparse Masked Modeling for Vision Decoding
— AK (@_akhaliq) November 15, 2022
abs: https://t.co/C6Dq7bX9VK
project page: https://t.co/6nTi6yb8xN
github: https://t.co/mnfKG8Nhj2 pic.twitter.com/DSxii6Exim
Fast Text-Conditional Discrete Denoising on Vector-Quantized Latent Spaces
— AK (@_akhaliq) November 15, 2022
abs: https://t.co/iX3hg5HOVn
github: https://t.co/TPwZNSJgaD
t2i model requiring less than 10 steps to sample high-fidelity images pic.twitter.com/veHSlEmkaG
Highly recommend this talk by @marksaroufim, "Dealing with Career Stagnation: my Machine Learning Story"
— Hamel Husain (@HamelHusain) November 14, 2022
This talk is amazingly relevant again today: https://t.co/H0VqhgZfim
DreamArtist: Towards Controllable One-Shot Text-to-Image Generation via Contrastive Prompt-Tuning with @Gradio SD Web UI
— AK (@_akhaliq) November 14, 2022
github: https://t.co/kPlCKwQNtZ pic.twitter.com/e4VryGVdGQ
OneFormer: One Transformer to Rule Universal Image Segmentation
— AK (@_akhaliq) November 14, 2022
abs: https://t.co/nZHKqesjBD
github: https://t.co/nQU7CL2PuY
propose OneFormer, a universal image segmentation framework that unifies segmentation with a multi-task train-once design pic.twitter.com/xlKgtF7Zcr
There's always something cringe on Twitter, here's a useful one!
— Jason Weston (@jaseweston) November 14, 2022
🚨 new paper 🚨
The CRINGE Loss: Learning what language not to model
Train your LM to not generate bad sequences.
Shows improvements on three tasks (safety, contradictions, open dialogue).https://t.co/yiAzzYQcbV pic.twitter.com/ZMmm31Xdil
You can also use embetter with computer vision models.
— Sebastian Raschka 📚 (@rasbt) November 11, 2022
Here's an example training a scikit-learn logistic regression model on MobileNet embeddings: https://t.co/TWXt86Fctn https://t.co/WLBNU0yUwf pic.twitter.com/JE7C8li7vn
