Build custom tract properties for nemo subnet.
build_preprocessor_export_params(asr_model, inference_target, *, nemo: InjectedNemoModule = INJECTED, axis_registry=None) -> T.Iterator[ExportParameters]
Build export parameters for the preprocessor of a NeMo ASR model.
export_nemo_asr_model(asr_model, inference_target, export_dir: Path, compress_registry: str, compress_method: T.Optional[str] = None, skip_preprocessor: bool = False, split_joint_decoder: bool = False, extra_cfg: T.Optional[T.Dict[str, T.Any]] = None, float_dtype: T.Optional[torch.dtype] = None, remove_unused_inputs: bool = True, dump_checked_io: bool = False, only_subnets: T.Optional[T.Collection[str]] = None, *, omegaconf: InjectedOmegaConfModule = INJECTED, axis_registry=None, **kwargs)
Export a generic NeMo ASR model to NNEF format using TractNNEF.
export_nemo_from_model(*, model, target: InferenceTarget, export_dir: Path, axis_reg: T.Optional[AxisSymbolRegistry] = None, cfg: NemoExportConfig) -> None
Export a NeMo ASR model using a structured configuration.
This is the public programmatic API. Handles dtype preparation
(float16 / mixed) internally so callers don't need to
apply model.half() or WrapPreprocessorCast themselves.
exportable_nemo_net(output_name, model, input_example, use_dynamo=False, batch_size: int = 1, float_dtype: T.Optional[torch.dtype] = None, *, nemo: InjectedNemoModule = INJECTED, pytorch_lightning: InjectedLightningModule = INJECTED)
Context manager to follow export way of nemo models.
It prepare model by switching mode to eval, disabling typechecks and wrapping forward method for tracing by PyTorch export tools.
Mostly borrowed from nemo codebase logic (with more modularity). see: nemo.core.classes.Exportable._export
Yield
ExportContext with input_example, output_example and dynamic_axes ready for export.
iter_decoder_joint_subnets(subnet, input_example, ctx_dynamic_axes, *, batch_size: int, remove_unused_inputs: bool, split_joint_decoder: bool)
Yield export tuples for the decoder_joint case.
- If split_joint_decoder is True: yields separate decoder and joint entries with their own input_examples and dynamic axes.
- Otherwise: optionally remove unused inputs, fix batch size on the input example, validate arity, and yield a single decoder_joint entry using the context-provided dynamic axes.
iter_export_params_for_generic_nemo_asr_model(asr_model, inference_target, skip_preprocessor: bool = False, split_joint_decoder: bool = False, remove_unused_inputs: bool = True, float_dtype: T.Optional[torch.dtype] = None, only_subnets: T.Optional[T.Collection[str]] = None, axis_registry=None) -> T.Iterator[ExportParameters]
Iterator over export parameters for a generic NeMo ASR model.
iter_nemo_model_subnets(model, input_example=None, float_dtype: T.Optional[torch.dtype] = None, split_joint_decoder: bool = False, remove_unused_inputs: bool = True, apply_sequential_examples: bool = False, batch_size: int = 3, only_subnets: T.Optional[T.Collection[str]] = None)
Iterator over exportable subnets of a nemo model.