Can CEBRA use 2d or 3d position as label?

Question

cnwangy opened this issue a year ago · 3 comments

Can CEBRA use 2d or 3d position as label? If I change the label to 2d delta, an error will be thrown

RuntimeError: The size of tensor a (2) must match the size of tensor b (4) at non-singleton dimension 1

Answer 1 · 2023-09-09T15:14:06.000Z

Hi @cnwangy thanks for your question. Could you provide us with example code and failed output, so we can better understand your query, thanks!

Answer 2 · 2023-09-09T15:24:17.000Z

Hi @cnwangy , for now I would recommend to switch to the time_delta distribution, then you can use higher dimensional outputs.

We might add support for higher dimensional labels when using the delta distribution in the future, right now this is not supported.

Answer 3 · 2023-09-13T01:55:02.000Z

Thanks a lot! I will give a try.