$ torch-model-archiver -h
usage: torch-model-archiver [-h] --model-name MODEL_NAME  --version MODEL_VERSION_NUMBER
                      --model-file MODEL_FILE_PATH --serialized-file MODEL_SERIALIZED_PATH
                      --handler HANDLER [--runtime {python,python3}]
                      [--export-path EXPORT_PATH] [-f] [--requirements-file] [--config-file]

Model Archiver Tool

optional arguments:
  -h, --help            show this help message and exit
  --model-name MODEL_NAME
                        Exported model name. Exported file will be named as
                        model-name.mar and saved in current working directory
                        if no --export-path is specified, else it will be
                        saved under the export path
  --serialized-file SERIALIZED_FILE
                        Path to .pt or .pth file containing state_dict in
                        case of eager mode or an executable ScriptModule
                        in case of TorchScript.
  --model-file MODEL_FILE
                        Path to python file containing model architecture.
                        This parameter is mandatory for eager mode models.
                        The model architecture file must contain only one
                        class definition extended from torch.nn.Module.
  --handler HANDLER     TorchServe's default handler name  or handler python
                        file path to handle custom TorchServe inference logic.
  --extra-files EXTRA_FILES
                        Comma separated path to extra dependency files.
  --runtime {python,python3}
                        The runtime specifies which language to run your
                        inference code on. The default runtime is
                        RuntimeType.PYTHON. At the present moment we support
                        the following runtimes python, python3
  --export-path EXPORT_PATH
                        Path where the exported .mar file will be saved. This
                        is an optional parameter. If --export-path is not
                        specified, the file will be saved in the current
                        working directory.
  --archive-format {tgz, no-archive, zip-store, default}
                        The format in which the model artifacts are archived.
                        "tgz": This creates the model-archive in <model-name>.tar.gz format.
                        If platform hosting requires model-artifacts to be in ".tar.gz"
                        use this option.
                        "no-archive": This option creates an non-archived version of model artifacts
                        at "export-path/{model-name}" location. As a result of this choice,
                        MANIFEST file will be created at "export-path/{model-name}" location
                        without archiving these model files
                        "zip-store": This creates the model-archive in <model-name>.mar format
                        but will skip deflating the files to speed up creation. Mainly used
                        for testing purposes
                        "default": This creates the model-archive in <model-name>.mar format.
                        This is the default archiving format. Models archived in this format
                        will be readily hostable on TorchServe.
  -f, --force           When the -f or --force flag is specified, an existing
                        .mar file with same name as that provided in --model-
                        name in the path specified by --export-path will
                        overwritten
  -v, --version         Model's version.
  -r, --requirements-file
                        Path to requirements.txt file containing a list of model specific python
                        packages to be installed by TorchServe for seamless model serving.
  -c, --config-file         Path to a model config yaml file.

# TS frontend parameters
# See all supported parameters: https://github.com/pytorch/serve/blob/master/frontend/archive/src/main/java/org/pytorch/serve/archive/model/ModelConfig.java#L14
minWorkers: 1 # default: #CPU or #GPU
maxWorkers: 1 # default: #CPU or #GPU
batchSize: 1 # default: 1
maxBatchDelay: 100 # default: 100 msec
responseTimeout: 120 # default: 120 sec
deviceType: cpu # cpu, gpu, neuron
deviceIds: [0,1,2,3] # gpu device ids allocated to this model.
parallelType: pp # pp: pipeline parallel; pptp: tensor+pipeline parallel. Default: empty
useVenv: Create python virtual environment when using python backend to install model dependencies
         (if enabled globally using install_py_dep_per_model=true) and run workers for model loading
         and inference. Note that, although creation of virtual environment adds a latency overhead
         (approx. 2 to 3 seconds) during model load and disk space overhead (approx. 25M), overall
         it can speed up load time and reduce disk utilization for models with custom dependencies
         since it enables reusing custom packages(specified in requirements.txt) and their
         supported dependencies that are already available in the base python environment.

# See torchrun parameters: https://pytorch.org/docs/stable/elastic/run.html
torchrun:
  nproc-per-node: 2

# TS backend parameters
pippy:
  rpc_timeout: 1800
  pp_group_size: 4 # pipeline parallel size, tp_group_size = world size / pp_group_size