Explore >> Select a destination
|
You are here 
|
www.lesswrong.com | ||
| | | | |
iclr-blogposts.github.io
|
|
| | | | | Diffusion Models, a new generative model family, have taken the world by storm after the seminal paper by Ho et al. [2020]. While diffusion models are often described as a probabilistic Markov Chains, their underlying principle is based on the decade-old theory of Stochastic Differential Equations (SDE), as found out later by Song et al. [2021]. In this article, we will go back and revisit the 'fundamental ingredients' behind the SDE formulation and show how the idea can be 'shaped' to get to the modern form of Score-based Diffusion Models. We'll start from the very definition of the 'score', how it was used in the context of generative modeling, how we achieve the necessary theoretical guarantees and how the critical design choices were made to finally arri... | |
| | | | |
resources.paperdigest.org
|
|
| | | | | The International Conference on Machine Learning (ICML) is one of the top machine learning conferences in the world. Paper Digest Team analyzes all papers published on ICML in the past years, and presents the 15 most influential papers for each year. This ranking list is automatically constructed ba | |
| | | | |
thesephist.com
|
|
| | | | | [AI summary] The text provides an in-depth overview of research on sparse autoencoders (SAEs) applied to embeddings for automated interpretability. It discusses methods for analyzing and manipulating embeddings, including feature extraction, gradient-based optimization, and visualization tools. The work emphasizes the importance of understanding model representations to improve human-computer interaction with information systems. Key components include: 1) Automated interpretability prompts for generating feature labels, 2) Feature gradients implementation for optimizing embeddings to match desired feature dictionaries, and 3) Visualizations of feature spaces and embedding transformations. The text also includes FAQs addressing the use of embeddings over lan... | |
| | | | |
lilianweng.github.io
|
|
| | | So far, I've written about two types of generative models, GAN and VAE. Neither of them explicitly learns the probability density function of real data, $p(\mathbf{x})$ (where $\mathbf{x} \in \mathcal{D}$) - because it is really hard! Taking the generative model with latent variables as an example, $p(\mathbf{x}) = \int p(\mathbf{x}\vert\mathbf{z})p(\mathbf{z})d\mathbf{z}$ can hardly be calculated as it is intractable to go through all possible values of the latent code $\mathbf{z}$. | ||
