Learning the integral of a diffusion model
A deep dive on flow maps.
Sander Dieleman
I write about machine learning, deep learning, music information retrieval, recommender systems, generative models and more.
Latest posts
Generative modelling in latent space
Latent representations for generative models.
Diffusion is spectral autoregression
A deep dive into spectral analysis of diffusion models of images, revealing how they implicitly perform a form of autoregression in the frequency domain.
Noise schedules considered harmful
The noise schedule is a key design parameter for diffusion models. Unfortunately it is a superfluous abstraction that entangles several different model aspects. Do we really need it?
The paradox of diffusion distillation
Thoughts on the tension between iterative refinement as the thing that makes diffusion models work, and our continual attempts to make it _less_ iterative.
The geometry of diffusion guidance
More thoughts on diffusion guidance, with a focus on its geometry in the input space.
Perspectives on diffusion
Perspectives on diffusion, or how diffusion models are autoencoders, deep latent variable models, score function predictors, reverse SDE solvers, flow-based models, RNNs, and autoregressive models, all at once!
Diffusion language models
Diffusion models have completely taken over generative modelling of perceptual signals -- why is autoregression still the name of the game for language modelling? Can we do anything about that?
Guidance: a cheat code for diffusion models
A quick post with some thoughts on diffusion guidance
Diffusion models are autoencoders
Diffusion models have become very popular over the last two years. There is an underappreciated link between diffusion models and autoencoders.