Patrick Kidger

That's quite a lot of code! Try reducing things down to a minimal (10-20 line) example of your issue. Additionally, providing a more thorough description of the error (tracebacks etc.)...

So the issue here is that you only have one training sample. The usual pattern in supervised learning is to have lots of training samples, and train the model on...

That looks about right, yes. Seq2seq models aren't that difficult, though! Completely untested example: ``` class Seq2Seq(torch.nn.Module): def __init__(self, input_size, hidden_size, output_size): self.encoder = torch.nn.GRU(input_size, hidden_size) self.decoder = torch.nn.GRU(1, hidden_size)...

This particular example takes an input of some length, and produces an output of the same length. (A simple choice just for illustrative purposes.) If you only want a length-1...

@he-ritter - you're completely correct that the initial z0 shouldn't be all zeros, and should instead be a function (e.g. a small MLP) of the first value of the time...

No. That remains unchanged: you would use the final state of the CDE as the initial condition of a neural ODE. I suggest familiarising yourself with how it would work...

Seq2seq is possible, e.g. just use the final value of a CDE as the initial condition of an ODE. You can return multiple values from cdeint by passing a vector...

Yep, that looks good to me, with one exception: you want `X.grid_points` rather than `X._t`. Incidentally you may find a variety of other resources interesting: - This paper on interpolation...

Have a read of Section 3.2.1.3 of [On Neural Differential Equations](https://arxiv.org/abs/2202.02435). The rest of Chapter 3 might also be helpful for context.

Nothing seems suspicious unfortunately; the CPU isn't being overloaded etc. Yes, the tmux comparison is using the same terminal implementation.