MMDialog: A Large-scale Multi-turn Dialogue Dataset Towards Multi-modal Open-domain Conversation
— AK (@_akhaliq) November 11, 2022
abs: https://t.co/dYJRDP1xCU
github: https://t.co/j95Y0of9GJ pic.twitter.com/DrH4rp5EKc
MMDialog: A Large-scale Multi-turn Dialogue Dataset Towards Multi-modal Open-domain Conversation
— AK (@_akhaliq) November 11, 2022
abs: https://t.co/dYJRDP1xCU
github: https://t.co/j95Y0of9GJ pic.twitter.com/DrH4rp5EKc
What the DAAM: Interpreting Stable Diffusion Using Cross Attention
— AK (@_akhaliq) November 11, 2022
abs: https://t.co/Y3vYiZRcID
github: https://t.co/OUQPlQ5a5F@gradio demo: https://t.co/Feu06pQOix pic.twitter.com/PLBwanJOnv
ZerO Initialization: Initializing Neural Networks with only Zeros and Ones
— hardmaru (@hardmaru) November 11, 2022
A fully deterministic initialization scheme which sets the weights to only 0s and 1s can achieve SOTA on various datasets including ImageNet. Maybe random weights are unnecessary.https://t.co/t6u3S6Dj71 pic.twitter.com/XgszDvat6T
StyleNAT: Giving Each Head a New Perspective
— AK (@_akhaliq) November 11, 2022
abs: https://t.co/3MTb1Nwtqn
github: https://t.co/EmY1eskFlP pic.twitter.com/pzQ8QWnk5Y
Creative Writing with an AI-Powered Writing Assistant: Perspectives from Professional Writers
— AK (@_akhaliq) November 10, 2022
abs: https://t.co/hcNMnv39df pic.twitter.com/XzEYOUgA8U
"Data Models for Dataset Drift Controls in Machine Learning With Images"
— Bojan Tunguz (@tunguz) November 9, 2022
Paper: https://t.co/uYxu74SIeg
Code: https://t.co/zpG4oo7bBx
Dataset: https://t.co/ohrAEQA1nE#MachineLearning #DeepLearning #ArtificialIntelligence #ML #DL #AI
2/2
How Much Does Attention Actually Attend? Questioning the Importance of Attention in Pretrained Transformers
— AK (@_akhaliq) November 8, 2022
abs: https://t.co/1jfwlUlsHt pic.twitter.com/BdUs7uMnSx
Rickrolling the Artist: Injecting Invisible Backdoors into Text-Guided Image Generation Models
— AK (@_akhaliq) November 7, 2022
abs: https://t.co/GaRXISObfs
github: https://t.co/l2mroWbLwT pic.twitter.com/CIz4kOW9Bx
Large Language Models Are Human-Level Prompt Engineers
— AK (@_akhaliq) November 4, 2022
abs: https://t.co/eqNfe2o4JQ
project page: https://t.co/tusmJehApu pic.twitter.com/Vw7UU36XJF
eDiffi: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers
— AK (@_akhaliq) November 3, 2022
abs: https://t.co/QaJfNzXeLL
project page: https://t.co/UrYmzGek8C
sota t2i diffusion model that consists of a base diffusion model and two super-resolution modules, 1024 × 1024 high-definition outputs pic.twitter.com/VN9qWla54v
On the detection of synthetic images generated by diffusion models
— AK (@_akhaliq) November 3, 2022
abs: https://t.co/TcyViS0RPH pic.twitter.com/1igXfAenCm
Text-Only Training for Image Captioning using Noise-Injected CLIP
— AK (@_akhaliq) November 2, 2022
abs: https://t.co/g4ZERgWtyW
github: https://t.co/HIpuHqCECx pic.twitter.com/Kxl4vHWHqx