Run a model operation.
By default Guild will try to run
OPERATION for the default model
defined in the current project.
MODEL is specified, Guild will use it instead of the default
OPERATION may alternatively be a Python script. In this case any
current project is ignored and the script is run directly. Options
in the format
‑‑NAME=VAL can be passed to the script using
flags (see below).
[MODEL]:OPERATION may be omitted if
‑‑rerun is specified, in
which case the operation used in
RUN will be used.
FLAG values in the form
‑‑rerun is specified, the operation and flags used in
will be applied to the new operation. You may add or redefine
flags in the new operation. You may also use an alternative
operation, in which case only the flag values from
RUN will be
RUN must be a run ID or unique run ID prefix or the
0, which indicates the latest run.
‑‑restart is specified, the specified
RUN is restarted
in-place using its operation and flags. Unlike rerun, restart does
not create a new run, but instead reuses the run directory of
RUN. Like a rerun, a restart may specify a different operation
and additional flags and may use
0 for the value of
restart the latest run.
‑‑run‑dir may not be used with
‑‑restart may not both be used.
Alternate run directory
To run an operation outside of Guild's run management facility,
‑‑stage to specify an alternative run
directory. These options are useful when developing or debugging
an operation. Use
‑‑stage to prepare a run directory for an
operation without running the operation itself. This is useful
when you want to verify dependency resolution and pre-processing
or manually run an operation in a prepared directory.
NOTE: Runs started with
‑‑run‑dir are not visible to Guild
and will not appear in run listings.
Controlling visible GPU devices
By default, operations have access to all available GPU
devices. To limit the GPU devices available to a run, use
For example, to limit visible GPU devices to
guild run ‑‑gpus 0,1 ...
To disable all available GPUs, use
‑‑no‑gpus are used to construct the
CUDA_VISIBLE_DEVICES environment variable used for the run
CUDA_VISIBLE_DEVICES is set, using either of these
options will cause it to be redefined for the run.
|-l, --label LABEL||
Set a label for the run.
|-d, --run-dir DIR||
Use alternative run directory DIR. Cannot be used with --stage.
Stage an operation in DIR but do not run. Cannot be used with --run-dir.
Use the operation and flags from RUN. Flags may be added or redefined in this operation. Cannot be used with --restart.
|-r, --remote REMOTE||
Run the operation remotely.
Restart RUN in-place without creating a new run. Cannot be used with --rerun or --run-dir.
A comma separated list of plugin names to disable. Use 'all' to disable all plugins.
Limit availabe GPUs to DEVICES, a comma separated list of device IDs. By default all GPUs are available. Cannot beused with --no-gpus.
Disable GPUs for run. Cannot be used with --gpu.
Do not prompt before running operation.
Accept all flag assignments, even for undefined or invalid flags.
Stop operation after N minutes.
Run only if there is not an available matching run. A matching run is of the same operation with the same flag values that is not stopped due to an error.
Run operation in background. PIDFILE must be a path to a file where the background process ID is written.
Don't wait for a remote operation to complete. Ignored if run is local.
Enter the Python debugger at the operation entry point.
Do not show output.
Show operation command and exit.
Show operation environment and exit.
Show model help and exit.
Show operation help and exit.
|--help||Show command help and exit.|