RESEARCH - AI Dynamics

I'm not sure about that. E.g. tricking the ZeroGPT algo means actually writing more natural text (so it'll be better, not worse), and tricking the watermarking means not being limited by the constraints of the watermarking algo.

→ View original post on X — @jeremyphoward

6 January 2023

Accessible Explanation of Diffusion Models Through Physical Ink Process

By

@animaanandkumar

–

06 January 2023 0h56

I love the accessible explanation of diffusion models through the physical process of ink diffusing through water and following that trajectory and reversing it.

→ View original post on X — @animaanandkumar

6 January 2023

Cognition in LLMs: Associative Memory vs. Functional Awareness

By

@goodside

–

06 January 2023 0h33

That’s a good point. What humans see as cognition in LLMs is often just good associative memory, or (generously) intuition. A model that can echo “I am aware” is not necessarily, but in larger LLMs more of the functional implications of awareness begin to apply.

→ View original post on X — @goodside

6 January 2023

Generative Models Potential for Improving Predictive Tasks

By

@animaanandkumar

–

06 January 2023 0h32

I talk about the enormous potential of generative models to improve downstream predictive tasks in this excellent article by @anilananth

→ View original post on X — @animaanandkumar

6 January 2023

AI-generated content detection methods easily circumvented

By

@dotcsv

–

05 January 2023 21h34

Me reitero en lo dicho, cualquier opción de detectar contenido generado por IA actualmente puede ser fácilmente contratado.

→ View original post on X — @dotcsv

5 January 2023

Training People-Diffusion Model for Image Generation

By

@levelsio

–

05 January 2023 21h08

Progress on training my own model called people-diffusion: 1) New Year celebrations in China
2) Protests in Brazil
3) Indigenous Bolivian
4) Woman at Times Square

→ View original post on X — @levelsio

5 January 2023