The Diffusion Models Course is an open source technical course on diffusion models! The first first unit was released today
RESEARCH
-
Free Diffusion Models Course Launch Event with Leading AI Experts
By
–
To inspire you for our just-released Diffusion Models Course with @johnowhitaker we are excited to share the free online event with @hardmaru
, @deviparikh
, @Buntworthy
, @robrombach
, @pess_r and @multimodalart on Nov 30th at 18h CET Register here: https://
huggingface.us17.list-manage.com/subscribe?u=7f
57e683fa28b51bfc493d048&id=ef963b4162
… -
NeurIPS Authors Overestimate Acceptance Chances Dramatically
By
–
Figure 1b) shows >= 90% of #NeurIPS2022 submissions thought they had a >= 25% chance of getting accepted. But only ~25% were accepted. Does this mean:
a) Authors overrate their work or b) Reviewers underrate most papers? Source: https://
blog.ml.cmu.edu/2022/11/22/neu
rips2021-author-perception-experiment/
… https://
cs.cmu.edu/~nihars/prepri
nts/NeurIPS_2021_Author_Perception_Experiment.pdf
… -
Stable Diffusion 2 Improves Safety Filter with SFW Focus
By
–
was catching up on this reading and noticed that the #NeurIPS2022 paper on “Red-Teaming the Stable Diffusion Safety Filter” is already out of date thanks to #StableDiffusion2 SD2 becoming a SFW "foundational txt2img" model means less spurious NSFW triggers! behold, dolphins!
-
Deep Learning Bayesian Methods Reveal Third Archaic Human Introgression
By
–
“Approximate Bayesian computation with deep learning supports a third archaic introgression in Asia and Oceania” – fascinating application.
-
Smart Actions and Active Intelligence for Data Analytics
By
–
Inform #Smart Actions in a Constantly Changing World. Honour to #write Issue 2 #ActiveIntelligence Executive Insights with @FryRSquared @tdav & more! #magazine https://
qlik.com/us/active-inte
lligence/magazine/issue-2?utm_campaign=7013z000001oUkJAAU&utm_content=ActiveMagazineSally&utm_medium=social&utm_source=Twitter&utm_team=COM&utm_mpt_id=CGCDLK
… #data #Analytics #100DaysOfCode #tech #skills #DevOps #AI @Qlik_UK #innovation @dinisguarda -
Pre-trained Language Models Essential for Agent Communication
By
–
If AZ cannot rely on a pre-trained LM to communicate, it is impossible (there is no reason the model learns with self play to communicate with valid sentences). Other issue will of course be the infinite action space (which we also had in theorem proving) but this is manageable
-
Powderworld: Understanding Generalization via Rich Task Distributions
By
–
Powderworld: A Platform for Understanding Generalization via Rich Task Distributions Kevin Frans, Phillip Isola : https://
arxiv.org/abs/2211.13051 #ArtificialIntelligence #DeepLearning #MachineLearning -
Rigor vs Impact: The Inverse Hierarchy of Scientific Disciplines
By
–
In the pecking order of rigor, mathematicians look down on physicists, which look down on engineers, which look down on computer scientists. In the pecking order of impact, it's the other way around.