zamba

zamba copied to clipboard

Documentation for Windows users

1

While it is not officially supported for maintenance reasons, it is now possible to run `zamba` on a windows machine. Add the small changes that are necessary and documentation to...

pjbull

This step is super slow ``` Checking all 245,944 filepaths exist. Can take up to a minute for every couple thousand files. ``` Looking it utilization while it's running, `goofys`...

AllenDowney

The current implementation has an extra level of nesting that doesn't seem necessary. Also, it stores `cache_dir` and `cleanup_cache` as attributes, but these options are also provided later in the...

AllenDowney

This page: https://zamba.drivendata.org/docs/stable/train-tutorial/ Contains some misleading/confusing information about the options, and some copy-pasted text that's actually about prediction.

AllenDowney

Before training, check user-provided train_labels to make sure it contains a validation split

2

Currently, if the user does not define splits, we generate random splits and check them. But if the use provides the splits, we don't do any checking. If the user-provided...

AllenDowney

Workflow failed: [tests #410](https://github.com/drivendataorg/zamba/actions/runs/2883514629)

github-actions[bot]

When this issue is resolved: https://github.com/mkdocstrings/mkdocstrings/issues/463 We should be able to undo the changes in this PR: https://github.com/drivendataorg/zamba/pull/207

AllenDowney

Make the signature of `load_video_frames` stricter

1

Currently `load_video_frames` can take a `VideoLoaderConfig` object as a parameter or, if it's missing, it gathers the kwargs and makes a `VideoLoaderConfig`. This permissiveness complicates the implementation if `npy_cache` and...

AllenDowney

Add basic depth module to zamba

4

### Not ready to merge into master The currently functionality is basic and geared towards what is needed to run a full pipeline based on the megadetector output. Before adding...

klwetstone

Set split proportions to None if split column is provided

2

If a user just provides a split column, they'll get the default split proportions. This leads to the message ``` Labels contains split column yet split_proportions are also provided. Split...

ejm714