Isaac Chung

@quest-bot embark

@quest-bot loot #866

@AzulGarza Yes that makes sense, thanks for the feedback. I've updated the PR description and pushed a new commit that reflect the changes requested.

Thanks @jmoralez . For this test, do you have a preference on where it lives? e.g. a new cell in `nbs/core.ipynb`?

@jmoralez thanks again. See if this aligns with what we had discussed above.

I wasn't able to run it before as my system has more than 1 GPU. I ran `export CUDA_VISIBLE_DEVICES=0` and it seems to have been resolved. Running the test in...

Makes sense. I updated the test to not include the last h periods. What do you mean by "generate the indices starting from the back"?

I'm confused as to 1) whether the last cutoff is needed in addition to the first one, and 2) what we want to keep in that example, where n_window =...

@jmoralez will do first thing when I'm back online in 2 weeks! 👍

@jmoralez I found why the test is failing. The actual cutoffs took into consideration of nf.uids (n_series = 2). Fix is in the newest commit.