search

DiffEqML / torchdyn

A PyTorch library entirely dedicated to neural differential equations, implicit models and related numerical methods
https://torchdyn.org
Apache License 2.0
1.33k stars 124 forks source link

Quickstrat desc possible mistake? #208

Closed stefanosIoannou closed 5 months ago

stefanosIoannou commented 5 months ago

Fairly new to NeuralODE, but from what I can gather a depth-invariant NeuralODE do not depend explicitly on time and not the state. So I changed the definition of depth-invariant NeuralODE from:

"This Neural ODE model is of depth-invariant type as neither f explicitly depend on s nor the parameters..." to "This Neural ODE model is of depth-invariant type as neither f explicitly depend on t nor the parameters..."

in the quickstart tutorial

Lmk if I messed up :]

Zymrael commented 5 months ago

We used s instead of t since often in neural differential equations the integration variable denotes the "depth" of the model, rather than time (especially when used in tasks like classification). But they're interchangeable.