Hi,
I met an unexpected output using LISBET in the self-supervised mode.
I want to train it on my data and the temporal axis was in secondes.
The model did the training but was not converging at all.
I discovered that it was because the expected time scale is in frame (so integer).
I do not know the impact in the code of using other scale.
On top of that, what will happen if the time scale is not continous ?
Hi,
I met an unexpected output using LISBET in the self-supervised mode.
I want to train it on my data and the temporal axis was in secondes.
The model did the training but was not converging at all.
I discovered that it was because the expected time scale is in frame (so integer).
I do not know the impact in the code of using other scale.
On top of that, what will happen if the time scale is not continous ?