evolve upload

Upload results from another machine or runner. Evolve imports their trials, rewards, and traces as a completed job.

evolve upload ./completed-job -d deepswe@1.1

This example links results for DeepSWE tasks to deepswe@1.1. Replace the dataset with the version used for your run. The source must use the supported job layout; a local directory needs config.json and result.json at its root. For an SDK session log, use the packaging walkthrough first.

Choose a source

Directory
Archive
Public URL

evolve upload ./completed-job

evolve upload ./completed-job.tar.gz

evolve upload --from https://example.com/completed-job.tar.gz

Use one source per invocation. --from requires a public HTTPS URL.

Options

Option	Meaning
`-d`, `--dataset <name[@version]>`	Link imported trial tasks to a published dataset version by task name.
`--from <url>`	Have Evolve fetch a public job archive. Replaces the local path.
`--no-wait`	Return the import record without waiting for the completed job.

Dataset linking makes the stored task available to later analysis. Review the report for unlinked tasks and skipped trial folders.

Follow an import

By default, upload waits for the imported job. Use --no-wait to return after the upload is accepted:

evolve upload ./completed-job --no-wait

Copy the complete import ID from that output into $IMPORT_ID, then follow it:

evolve job imports --status RUNNING
evolve job import "$IMPORT_ID" --watch

With --json, the default command prints one final job document. With --no-wait --json, it prints the import record instead. A failed import returns exit code 1 when watched. Exit code 0 with --no-wait confirms acceptance; the import can still fail later.

Inspect the result

Use the imported job’s ID from the completed upload or job import output as $JOB_ID:

evolve job show "$JOB_ID"
evolve job trials "$JOB_ID"
evolve analyze "$JOB_ID" --watch

Uploaded costs are reported by the source archive, not metered execution on Evolve. Uploaded jobs can be analyzed, but cannot be resumed, retried, or regraded. Global options apply.

Start work

Inspect results

Manage resources

Choose a source

Options

Follow an import

Inspect the result