- Robotics
- RL
- CV
- formatting
- algorithms
- experiments
- math
•
•
•
•
•
•
-
Notes for Ajoint Methods
The adjoint method, introduced in the 2018 neural ODE paper, provides an efficient approach for computing gradients in neural ODEs. This article presents a detailed mathematical derivation of the adjoint method's core formulas and their applications.
-
Notes for Diffusion Models
Diffusion models are an important class of modern generative models. This article provides a brief introduction to distribution-based generative algorithms such as diffusion, SDE, and consistency models based on my personal understanding.