nat.plugins.eval.utils.output_uploader#
Attributes#
Classes#
Run custom scripts and upload evaluation outputs using the configured s3 |
Functions#
Module Contents#
- logger#
- S3_UPLOAD_INSTALL_HINT = 'S3 output upload requires optional dependencies that are not installed. Install with: uv pip...#
- _load_s3_upload_dependencies()#
- class OutputUploader(
- output_config: nat.data_models.evaluate_config.EvalOutputConfig,
- job_id: str | None = None,
Run custom scripts and upload evaluation outputs using the configured s3 credentials.
- output_config#
- _s3_client = None#
- job_id = None#
- property s3_config#
- async _upload_file(s3_client, bucket, s3_key, local_path, pbar)#
- async upload_directory()#
Upload the contents of the local output directory to the remote S3 bucket in parallel.
- run_custom_scripts()#
Run custom Python scripts defined in the EvalOutputConfig. Each script is run with its kwargs passed as command-line arguments. The output directory is passed as the first argument.