mim
mim copied to clipboard
add `model_info_local` argument to `download` function
Motivation
We want to parse model info from remote when using download function. Fixes https://github.com/open-mmlab/mim/issues/51
Modification
add model_info_local argument to download function.
BC-breaking (Optional)
Previous usage is not affected.
Thanks for your contribution. But why we need download a checkpoint from remote GitHub.
Thanks for your contribution. But why we need download a checkpoint from remote GitHub.
@zhouzaida because model_zoo.yml file may not be available at local.
got it
Got it. Please add an unittest. Another thing is that the config file can not be successfully obtained from remote because the config is parsed from the local repo.
Got it. Please add an unittest. Another thing is that the config file can not be obtained from remote because the config is parsed from the local repo.
@zhouzaida it is also an issue. wheel package of mmdetection does not include configs directory and mim gives an error when trying to load config from local. I have created a separate issue for that: https://github.com/open-mmlab/mim/issues/50
@zhouzaida added tests for model_info_local=False.
hi @fcakyon , the following lines should be moved. In addition, we download checkpoints and configs by default so we can provide two options to only download one of them https://github.com/open-mmlab/mim/blob/eb35cceb8411f3767ffa374a462cf19ccc083e50/mim/commands/download.py#L63-L71
Codecov Report
:exclamation: No coverage uploaded for pull request base (
main@1bb17cd). Click here to learn what that means. The diff coverage isn/a.
:exclamation: Current head eb35cce differs from pull request most recent head 74e8203. Consider uploading reports for the commit 74e8203 to get more accurate results
@@ Coverage Diff @@
## main #49 +/- ##
=======================================
Coverage ? 63.81%
=======================================
Files ? 21
Lines ? 1564
Branches ? 347
=======================================
Hits ? 998
Misses ? 431
Partials ? 135
| Flag | Coverage Δ | |
|---|---|---|
| unittests | 63.81% <0.00%> (?) |
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact),ø = not affected,? = missing dataPowered by Codecov. Last update 1bb17cd...74e8203. Read the comment docs.