convert.convert

convert_model_kernel

def convert_model_kernel(model: nn.Module, config: dict, export_path: str,
                         skip_exist: bool, argument_infos: Any,
                         logger: Any) -> dict

Convert model based on configuration settings.

Arguments:

• model nn.Module - PyTorch model to convert
• config dict - Conversion configuration dictionary
• export_path str - Base path for exporting converted model
• skip_exist bool - Whether to skip existing conversion files
• argument_infos Any - Input argument information for conversion
• logger Any - Logger object for tracking conversion progress

Returns:

• dict - Conversion summary with updated configuration

convert_model

def convert_model(model: nn.Module,
                  config: dict,
                  export_path: str,
                  skip_exist: bool,
                  data_loader: Union[Iterable[Union[Tuple[Tuple, Dict], Dict]],
                                     None] = None,
                  data_loader_post_process: Union[Callable, None] = None,
                  logger=None)

Convert model with preprocessing and postprocessing.

Arguments:

• model nn.Module - PyTorch model to convert
• config dict - Conversion configuration dictionary
• export_path str - Base path for exporting converted model
• skip_exist bool - Whether to skip existing conversion files
• data_loader Union[Iterable[Dict], None], optional - Data loader for preprocessing. Defaults to None.
• data_loader_post_process Union[Callable, None], optional - Postprocessing function for data loader. Defaults to None.
• logger optional - Logger object for tracking conversion progress. Defaults to None.

Returns:

• dict - Conversion configuration dictionary