Mr-Geekman

I'll try to explain the core of the issue. We have a wide dataframe `df`. We want to select a few columns: `[column_1, column_2]`: ```python res = df.loc[:, pd.IndexSlice[:, [column_1,...

More detailed results. Imagine we have a `df_wide` with segments: `["segment_2", "segment_1", "segment_0"]` and with features: `["target", "exog_2", "exog_1", "exog_0"]`. Calling `df_wide.loc[:, pd.IndexSlice[:, ["exog_1", "exog_2"]]]` gives order of columns: 1....

Waiting for implementation of inference mode.

Draft PR with merged master into inference: #979.

The script works fine on macOS 12.5.1 with Intel CPU.

Package `pytorch_forecasting` [was updated](https://github.com/jdb78/pytorch-forecasting/releases/tag/v1.0.0) recently. It looks like they probably solved the problem with `pytorch_lightning` there. But other packages requirements are very strict.

We aren't going to do it now, move it to "Specification" to revisit in the future.

It is now possible after completion of https://github.com/tinkoff-ai/etna/milestone/23

We are currently doing [AutoML track](https://github.com/tinkoff-ai/etna/milestone/19) and the chosen path is very similar to the suggested here.

Waiting for #886.