Victor Rodriguez-Fernandez

You can use `combine_split_data` as a way of merging separated datsets into a single X,y,splits triplet. In terms of meta-learning and customising the training loop...I'd rather look at the fastai's...

> I think option 3 makes more sense I wonder, why does it make more sense to pad at the beginning than at the end? > a sequence has 2...

Mmm ok I see the point, when you have to choose what to keep, you give priority to the future, because at the end, it's what matters the most generally...

oh yes sure, it's the same situation in both cases :) looks coherent to me!!

Wow, amazing work Ignacio!!! This is a huge addition to the library. > It has a state so you can fit on the train set, and then transform train, validation,...

this is great, thank you!

This is exciting Ignacio, thank you so much!!!

Hi, With `learn.model` you access the Pytorch model (instance of nn.Module) that has been trained in the learning process.

What you are seeing are the activation heatmaps for the two branches of the XCM: the 2D conv (first figure, shape=[1xvarsxseq_len]) and the 1D conv (second figure, shape=[1x1xseq_len]). What I'm...

exactly, in the second plot the y axis is irrelevant because the feature map represented is one dimensional (comes from a 1D conv).