Sequential models in DeepXDE

[dang-trinh156]

Dear Dr.Lulu and developer team and users,
I've scanned through the FAQ and search on this discussion session. All what I can find is this post:#1488.
Also, I've read the source code in Geometry, icbc, PDE, and they're all implemented on 2D-arrays (i.e. [batch_size or n_data, spatial-temporal features]). Here's an example:

The error function returns an array/tensor in 2D. Therefore, only FCNN architectures are used.

Question:
Have there been any examples, or snippets of code about the implementation of 3D inputs (i.e. [batch_size, sequence length, features]) for sequential models?

Thank you in advance!

[lululxvi]

No. Then you need to implement the network by yourself.