Stability AI unveiled Stable Diffusion 3, a new iteration of the company’s image generator: -Improvements in quality
-Multi-subject generation
-Enhanced spelling The model is backed by a new transformer architecture similar to OpenAI’s Sora.
RESEARCH
-
Stability AI Releases Stable Diffusion 3 with Transformer Architecture
By
–
-
Criticism of Pessimistic AI Discourse Calls for Hopeful Vision
By
–
Your messages spread hopelessness. I genuinely wish you would use your amazing brain and energy inspiring something akin to a positive future. You make me sad.
-
Defining Equitarism: Academic Framework Beyond Social Justice
By
–
It needs to be non-disparaging, ie be accepted by the adherents, but also delineate it from other perspectives (such as liberalism). "Social Justice" is probably not academic enough. Equitarism? The practice of rewriting history is "elimination of harmful stereotypes"
-
Google’s AI Model Receives Positive Reassessment From Community
By
–
I think you're all too hard on Google. This model is really solid.
-
Causal Reasoning: The Next Frontier in Machine Learning
By
–
Causal reasoning is a key next frontier in Machine Learning.
-
Political Bias in Scientific Hypothesis Validation: A Critical Analysis
By
–
Even in physics and biology, dialectic materialism was teleological and normative, and explicitly permitted the rejection and endorsement of scientific hypotheses on political grounds.
-
Defining Computing Systems Through Finite State Automata Formalism
By
–
The terms are semantically loaded and seem to ambiguous to easily define them via first principles. “Class of systems that can be described via finite state automata” seems more elegant
-
Data Scientist Contributes Chapter to Portuguese AI Ethics Book
By
–
I was honored to contribute a chapter to this book by @hevertonsa Heverton Anunciacao => "The Data Scientist and his Demons" (in Portuguese, "O Cientista de Dados e Seus Demônios") at https://
amzn.to/49KFAT2 My chapter is based on an article, and 2 talks here: 1) -
Converting Video Lectures into Written Content Using LLMs
By
–
Fun LLM challenge that I'm thinking about: take my 2h13m tokenizer video and translate the video into the format of a book chapter (or a blog post) on tokenization. Something like: 1. Whisper the video
2. Chop up into segments of aligned images and text
3. Prompt engineer an LLM -
Reflexion: Agents Learn to Reflect and Self-Correct
By
–
2/ Paper: Reflexion, by Shinn, @ashwingop
, @ShunyuYao12 , et. al. https://
arxiv.org/abs/2303.11366