Skip to content

dataset push

Pushes a new dataset to a FlexAI. Multiple files can be uploaded at once by using multiple instances of the --file flag, or by pointing the --file flag to a directory containing dataset files.

The Uploading Datasets guide provides more information on how to use the dataset push command to upload both local files and files from a Remote Storage Provider.

Terminal window
flexai dataset push <dataset_name> \
(( --file <source_path> ... | --file <source_path>=<destination_path> ...) | [( --storage-provider <storage_provider_name> --source-path <source_path> )])
Required

Name to be used to identify the Dataset while performing operations on it.

Examples
  • wikitext-2-raw-v1
-f , --file
<string><key=value>
Optional
Path

The local path to the dataset files on the host machine running the FlexAI CLI.

The Dataset will be mounted under /input/<dataset_name> at runtime by default.

Examples
  • --file tdata.tar.gz
Key Value Path Mapping

A mapping between a local resource and the remote destination path where it will be uploaded.

The syntax is <source_path>=<destination_path>, where <destination_path> is located under /input/.

Examples
  • --file tdata.tar.gz=dataset/tdata.tar.gz
Optional

The path to the bucket and file or directory to be pushed.

Syntax: <bucket_name>/<path>.

Used in together with the --storage-provider flag.

Examples
  • --source-path my-bucket/datasets/coffee-leaf-diseases--train
Optional