siddhant

Knowledge / Generative AI

Diffusion Models

Forward noising, denoising, score estimation, conditioning, and diffusion-based generation.

By Siddhant Krishna · Published 2026-10-06 · Updated 2026-10-06

01

Forward and Reverse Processes

A diffusion system defines a controlled forward process that progressively corrupts data and trains a model to approximate the reverse transformation.

q(x_t | x_{t-1})
p_θ(x_{t-1} | x_t)

02

Conditioning

Conditioning allows a generative process to be directed by text, class labels, images, or other signals. The conditioning mechanism determines how strongly external information influences the generated sample.

References

  1. Ho, Jain & Abbeel (2020).
    https://arxiv.org/abs/2006.11239

Related

Contact

Get in Touch

Want to chat? Just shoot me a dm with a direct question on twitter and I'll respond whenever I can. I will ignore all soliciting.