-
I have a large system of coupled ODEs (~280x16). The loss computation, which involves solving odes (I am using odeint from experimental), takes 163ms. The gradient step takes ~33 secs. It is a relativ…
ibulu updated
2 years ago
-
I am sure you are probably aware, but the basic examples like ode_demo and latent_ode are broken in the rewrite branch. The size of the tensor returned from odeint is different from the input, and thu…
-
New config at https://github.com/jiweiqi/CellBox.jl/blob/main/Test_Beeline/Curated/mCAD/config.yaml
The neural network part is realized by reloading the ode function.
-
Using an explanation/implementation of an ODEnet as shown in https://github.com/msurtsukov/neural-ode
Using [torchdiffeq], implement an ODEnet that accepts as input a series of data, and predicts a f…
-
Very interesting work. I want to implement ACA method, should I use odesolver_mem instead?
Is the odesolver for the naive Neural ODEs using auto differentiation?
Would you have an example using…
-
Hi!
Just wondering how the RNN could be mixed into the `ODEProblem`
In flux times, it seems a Recur layer need to be created. However there is already a `Recurrence` in Lux.jl
[Training of UDEs w…
-
### 🐛 Describe the bug
I am experiencing a problem on my M1 Pro MacBook with training a Neural ODE model.
My use-case is quite extensive and depends on multiple python files and some custom integr…
-
Hi, great work, and thanks for the code!
I was wondering if the following is possible.
I have a system of ODEs (two ODEs):
`x_state = f(x, t, theta)`
`0 = d {g(x, t, phi)} / dx`
Simply …
ojus1 updated
2 years ago
-
As far as I can tell at the moment, the solution just returns the solution at t=T. Is it possible to return a trajectory of values at intermediate times? e.g. https://docs.kidger.site/diffrax/api/save…
-
( [Project Plan](http://bit.ly/pdspplan) )
**Which deliverable?**
D7
**Which repos (e.g., tx-logging)?**
**Short name**
classifier guidance plugin (DOAC)
**Description**
create a 'guidan…