avala datasets upload CLI command, or the Python SDK. These paths create a private dataset in Avala-managed storage. You authenticate with Avala; you do not need your own bucket or AWS credentials.
Use the wizard for interactive uploads. Use the CLI or SDK for directory trees, large files, and scripted transfers.
Storage allowance and file limits
Your upload allowance belongs to the dataset’s owner: your personal account or an organization. It is shared across that owner’s managed datasets, rather than reset for each dataset or organization member. The configured default organization allowance is 100 GiB (107,374,182,400 bytes). Personal accounts and owners with a custom allowance can have different limits. Check the API for your owner’s actuallimit and used bytes; do not assume every account has a 100 GiB allowance or access to the same upload modes.
You can see usage in the wizard or query it programmatically. These examples check your personal allowance. For an organization, pass its UID as organization_uid to the SDK or as a query parameter to the same endpoint. Use that same organization for the upload.
used includes in-flight reservations as well as stored bytes. Remaining capacity is max(0, limit - used) at the time of the response. The API checks capacity again when admitting uploads, so another upload can consume that space before yours starts.
There is no universal 2 GB file limit. The API selects the available transfer mode and enforces its file limits and your owner’s remaining allowance. Supported managed uploads use a direct PUT or multipart transfer for large files. Legacy POST uploads have a separately configured single-file limit. Use the current SDK or CLI to handle the returned mode; multipart support does not bypass quota checks.
If you need a higher allowance, contact support@avala.ai. For files that should stay in your own bucket, use cloud storage integration instead of uploading a second copy.
Web wizard (drag-and-drop)
- Open Mission Control and click Add Dataset.
- Name the dataset, choose its owner and data type, and set labels.
- Pick Local files as the source.
- Drag files into the drop zone or click to browse. Select file types accepted for your chosen data type. Each row shows upload progress.
- Retry any failed files, then click Submit after every selected file has uploaded. Dataset creation starts server-side indexing; wait for processing to finish before starting annotation.
CLI (avala datasets upload)
Install the CLI extra:
AVALA_API_KEY in your environment. Then upload a directory:
--organization-uid "$AVALA_ORGANIZATION_UID" to use an organization’s allowance and create the dataset under that organization. Without it, the upload belongs to your personal account.
The CLI walks the source directory, transfers files using API-issued upload URLs, completes managed uploads, and creates the dataset. It handles PUT and multipart transfers when the API returns them, and supports legacy POST responses. --wait also waits for indexing.
Resume is enabled by default. After an interruption, rerun the same command with the original files and owner. The CLI retains local checkpoints and reconciles completed uploads and multipart parts with the server where supported. Keep those checkpoints until the operation succeeds.
Common flags
Quota errors
When an upload exceeds the owner’s allowance, the API returns HTTP 413 withused and limit, and the CLI reports the quota error. Earlier files may already have uploaded. Resolve the capacity issue before retrying with the same owner and saved state. Removing a dataset from a list does not necessarily reclaim its stored bytes immediately; contact support if reported usage does not match your expectations.