Open youyinnn opened 1 year ago
Plus one on that
Seconding that.
I'm now going to need to complete approx 1000 sequences of 3 clicks in order to gather the data I require. (My usecase: I need to download a large selection of the competition files, but omitting a selection that total ~ 40GB that I don't have the disk space for.)
There is a --page
argument (default 20) for kaggle datasets download
and it seems like only twenty files from each dir above are showing. Coincidence?
OK, for anyone that finds themselves wanting to do this, but without a solution, here is a work-around:
competition_files.csv
.
competition
to your desired competition.competition_files.csv
either manually or via the kaggle api.OK, for anyone that finds themselves wanting to do this, but without a solution, here is a work-around:
Run a version of this kernel to walk the competition directory and push it to
competition_files.csv
.
- Change the
competition
to your desired competition.- You have to have the competition data loaded in the script env for anything useful to come out.
- You can then download the
competition_files.csv
either manually or via the kaggle api.
An "easy" way to have the competition file content in the input directory is to launch the notebook from the competition page and then execute the code in the notebook provided by @jfcann For instance, for the competition, RSNA 2023 Abdominal Trauma Detection, navigate to the competition --> Code (https://www.kaggle.com/competitions/rsna-2023-abdominal-trauma-detection/code) --> New Notebook