Isaac Godfried

https://icml.cc/virtual/2023/poster/24923

Finish series_id loader

2

wss

Finish end-to-end example of running sweep locally or on Docker container

1

Most of our tutorials demonstrate running in a Jupyter notebook and not from CLI. We should create a tutorial that clearly shows how to run a sweep from a CLI.

add profiling using cprofile

6

The goal of this PR is add automatic optimization/time calculation metrics. Also maybe do some optimization of some the code and find out where it does not run efficiently.

We we don't plot a confidence interval there is no plotting for multiple targets. https://github.com/AIStream-Peelout/flow-forecast/blob/0edb0178d2f2a7f8c02c24e0320d016dea856882/flood_forecast/trainer.py#L66

Clarify batch size in tutorials

1

Currently it is unclear in a number which batch_size parameter actually impacts the batch_size. Is it the one in dataset_params or another one?

We could use a detailed end-to-end example of using an AutoEncoder to create representations of temporal data. This should likely be done on Kaggle then added to the flow tutorials...