floydhub / floyd-cli

Command line tool for FloydHub - the fastest way to build, train, and deploy deep learning models
https://floydhub.com
Apache License 2.0
156 stars 34 forks source link

add optional argument to `floyd data clone` to save outputs to target directory #193

Open cosmicBboy opened 6 years ago

cosmicBboy commented 6 years ago

I have the use case where I'd like to programmatically download outputs to my local machine so that I can analyze the outputs across many jobs. Currently, the CLI option saves to the current directory and it's not straightforward to just dump all the outputs in a target directory because the output.tar file is unzipped.

Seems like it would be a quick solution to add an argument to the data clone command to set untar=False, delete_after_untar=False so that I can just mv the tar file to the target directory and untar it there.