When you click New Dataset on the Datasets page, you will be asked to add a new dataset either from a local or external data source. You have three data provider options: VESSL, Amazon Simple Storage Service, and Google Cloud Storage.
Managed
S3
GCS
Local storage
When you select a VESSL dataset, you can upload data from the local
disk. To create a VESSL dataset:Enter dataset name
Enter Dataset Name.
Upload files
Click Upload Files.
You can retrieve a dataset from S3 by selecting Amazon Simple Storage
Service. To create a dataset from S3:Enter dataset name
Enter Dataset Name.
Enter bucket path
Enter Bucket Path.
You also have an option to retrieve a dataset from Google Cloud Storage.
To create a dataset from GCS:Enter dataset name
Enter Dataset Name.
Enter bucket path
Enter Bucket Path.
Create
Click the Create button.
If the dataset exists inside the cluster (e.g., NAS, host machine) and you
want to mount it only inside the cluster, you can select the Local
Storage option. In this case, VESSL only stores the location of the
dataset, and mounts the path when an experiment is created. VESSL supports
3 types of local mounts:
Since VESSL does not have access to the local dataset, you cannot browse
local dataset files on VESSL. A detailed integration guide is provided on each Create Dataset dialog.
