modal-dojo run

"# modal-dojo run\n\nInspect and manage runs.\n\n**Usage**:\n\n```bash\nmodal-dojo run [OPTIONS] COMMAND [ARGS]...\n```\n\n**Options**:\n\n* `--help`: Show this message and exit.\n\n**Commands**:\n\n* [`modal-dojo run get`](#modal-dojo-run-get): Show a run's status and top-level metadata.\n* [`modal-dojo run params`](#modal-dojo-run-params): Show a run's framework training recipe.\n* [`modal-dojo run logs`](#modal-dojo-run-logs): Show logs for a training run.\n* [`modal-dojo run trace`](#modal-dojo-run-trace): Download agent traces for a run.\n* [`modal-dojo run list`](#modal-dojo-run-list): List training runs and top-level metadata.\n\n## `modal-dojo run get`\n\nShow a run's status and top-level metadata.\n\n**Usage**:\n\n```bash\nmodal-dojo run get [OPTIONS] RUN_ID\n```\n\n**Options**:\n\n* `--verbose`: Include reward-over-time and rollout data.\n* `-j, --json`: Output machine-readable JSON.\n* `--help`: Show this message and exit.\n\n**Examples**:\n\n```bash\nmodal-dojo run get brave-falcon-3fa8\nmodal-dojo run get brave-falcon-3fa8 --verbose\n```\n\n## `modal-dojo run params`\n\nShow a run's framework training recipe.\n\n**Usage**:\n\n```bash\nmodal-dojo run params [OPTIONS] RUN_ID\n```\n\n**Options**:\n\n* `-j, --json`: Output machine-readable JSON.\n* `--help`: Show this message and exit.\n\n**Examples**:\n\n```bash\nmodal-dojo run params brave-falcon-3fa8\nmodal-dojo run params brave-falcon-3fa8 --json\n```\n\n## `modal-dojo run logs`\n\nShow logs for a training run.\n\n**Usage**:\n\n```bash\nmodal-dojo run logs [OPTIONS] RUN_ID\n```\n\n**Options**:\n\n* `-f, --follow`: Keep streaming new logs until interrupted or the run stops.\n* `--since START`: Show logs since a timestamp or relative time, such as 30m or 2h.\n* `--until END`: Show logs up to a timestamp or relative time, such as 30m or 2h.\n* `-n, --tail N`: Show at most N recent logs in the requested window. Default `100`; range `1` to `20000`.\n* `--search TEXT`: Filter by search text.\n* `-j, --json`: Output machine-readable JSON.\n* `--help`: Show this message and exit.\n\n**Examples**:\n\n```bash\nmodal-dojo run logs brave-falcon-3fa8 --follow\nmodal-dojo run logs brave-falcon-3fa8 --since 30m -j\n```\n\n## `modal-dojo run trace`\n\nDownload agent traces for a run.\n\n**Usage**:\n\n```bash\nmodal-dojo run trace [OPTIONS] RUN_ID\n```\n\n**Options**:\n\n* `--out DIR`: Output directory. Required.\n* `--step STEP`: Select steps by list or end-exclusive range, or omit to select all steps.\n* `--dry-run`: Report step/sample count and approximate size without downloading.\n* `-y, --yes, --force`: Run without pausing for confirmation.\n* `-j, --json`: Output machine-readable JSON.\n* `--help`: Show this message and exit.\n\n**Examples**:\n\n```bash\nmodal-dojo run trace brave-falcon-3fa8 --out ./traces --step 4-100:2\nmodal-dojo run trace brave-falcon-3fa8 --out ./traces --step 1,4,9 --dry-run\nmodal-dojo run trace brave-falcon-3fa8 --out ./traces --yes\n```\n\n## `modal-dojo run list`\n\nList training runs and top-level metadata.\n\nResults are sorted by most recently updated.\n\n**Usage**:\n\n```bash\nmodal-dojo run list [OPTIONS]\n```\n\n**Options**:\n\n* `--status STATUS`: Filter by status.\n* `--training-type TRAINING_TYPE`: Filter by type.\n* `--model MODEL`: Filter by model.\n* `--dataset DATASET`: Filter by dataset.\n* `--recipe RECIPE`: Filter by recipe.\n* `--group GROUP`: Filter by group.\n* `--since TIME`: Include runs created or updated since this timestamp or relative time.\n* `--limit N`: Maximum number of runs to return. Default `50`; minimum `1`.\n* `-j, --json`: Output machine-readable JSON.\n* `--help`: Show this message and exit.\n\n**Examples**:\n\n```bash\nmodal-dojo run list --status failed --since 24h\nmodal-dojo run list --status completed --group nightly-tau-bench -j\n```\n"

Inspect and manage runs.

Usage:

modal-dojo run [OPTIONS] COMMAND [ARGS]...

Options:

--help: Show this message and exit.

Commands:

modal-dojo run get: Show a run’s status and top-level metadata.
modal-dojo run params: Show a run’s framework training recipe.
modal-dojo run logs: Show logs for a training run.
modal-dojo run trace: Download agent traces for a run.
modal-dojo run list: List training runs and top-level metadata.

`modal-dojo run get`

Show a run’s status and top-level metadata.

Usage:

modal-dojo run get [OPTIONS] RUN_ID

Options:

--verbose: Include reward-over-time and rollout data.
-j, --json: Output machine-readable JSON.
--help: Show this message and exit.

Examples:

modal-dojo run get brave-falcon-3fa8
modal-dojo run get brave-falcon-3fa8 --verbose

`modal-dojo run params`

Show a run’s framework training recipe.

Usage:

modal-dojo run params [OPTIONS] RUN_ID

Options:

-j, --json: Output machine-readable JSON.
--help: Show this message and exit.

Examples:

modal-dojo run params brave-falcon-3fa8
modal-dojo run params brave-falcon-3fa8 --json

`modal-dojo run logs`

Show logs for a training run.

Usage:

modal-dojo run logs [OPTIONS] RUN_ID

Options:

-f, --follow: Keep streaming new logs until interrupted or the run stops.
--since START: Show logs since a timestamp or relative time, such as 30m or 2h.
--until END: Show logs up to a timestamp or relative time, such as 30m or 2h.
-n, --tail N: Show at most N recent logs in the requested window. Default 100; range 1 to 20000.
--search TEXT: Filter by search text.
-j, --json: Output machine-readable JSON.
--help: Show this message and exit.

Examples:

modal-dojo run logs brave-falcon-3fa8 --follow
modal-dojo run logs brave-falcon-3fa8 --since 30m -j

`modal-dojo run trace`

Download agent traces for a run.

Usage:

modal-dojo run trace [OPTIONS] RUN_ID

Options:

--out DIR: Output directory. Required.
--step STEP: Select steps by list or end-exclusive range, or omit to select all steps.
--dry-run: Report step/sample count and approximate size without downloading.
-y, --yes, --force: Run without pausing for confirmation.
-j, --json: Output machine-readable JSON.
--help: Show this message and exit.

Examples:

modal-dojo run trace brave-falcon-3fa8 --out ./traces --step 4-100:2
modal-dojo run trace brave-falcon-3fa8 --out ./traces --step 1,4,9 --dry-run
modal-dojo run trace brave-falcon-3fa8 --out ./traces --yes

`modal-dojo run list`

List training runs and top-level metadata.

Results are sorted by most recently updated.

Usage:

modal-dojo run list [OPTIONS]

Options:

--status STATUS: Filter by status.
--training-type TRAINING_TYPE: Filter by type.
--model MODEL: Filter by model.
--dataset DATASET: Filter by dataset.
--recipe RECIPE: Filter by recipe.
--group GROUP: Filter by group.
--since TIME: Include runs created or updated since this timestamp or relative time.
--limit N: Maximum number of runs to return. Default 50; minimum 1.
-j, --json: Output machine-readable JSON.
--help: Show this message and exit.

Examples:

modal-dojo run list --status failed --since 24h
modal-dojo run list --status completed --group nightly-tau-bench -j