qubee.frontend.Model_Dict Class Reference

qubee.frontend.Model_Dict Class Reference#

Mobilint SDK qb Compiler: qubee.frontend.Model_Dict Class Reference

Mobilint SDK qb Compiler v0.11.0.1

MCS002-EN

Wrapper around the Mobilint compiler to support compilation and inference workflows. More...

Inheritance diagram for qubee.frontend.Model_Dict:

Public Member Functions
	__init__ (self, model, backend="onnx", device="cpu", feed_dict=None, dynamic_axes=None, in_dformats=None, yolo_decode_include=False, **kwargs)
	Initialize the Mobilint compiler wrapper.
	compile (self, str calib_data_path=None, str save_path=None, int save_subgraph_type=0, output_subgraph_path="", str model_nickname=None, save_sample=False, use_random_calib=False, inference_scheme="single", cpu_offload=False, singlecore_compile=False, optimize_option=0, str sample_dtype="float", preprocess_dict=None, buffer_mode=1, quant_config=QuantConfig.default_config(), **kwargs)
	Compile the wrapped model into an MXQ artifact.

Wrapper around the Mobilint compiler to support compilation and inference workflows.

Capabilities include:

Definition at line 23 of file frontend.py.

qubee.frontend.Model_Dict.__init__	(		self,
			model,
			backend = "onnx",
			device = "cpu",
			feed_dict = None,
			dynamic_axes = None,
			in_dformats = None,
			yolo_decode_include = False,
		**	kwargs )

Initialize the Mobilint compiler wrapper.

Parameters

model	string or model instance. Model path or in-memory model to compile.
backend	string. Framework identifier for the model (for example "onnx"). Defaults to "onnx".
device	string. Target device for inference ("cpu" or "gpu"). Defaults to "cpu".
feed_dict	dict. Example inputs for shape resolution.
dynamic_axes	dict. Marks model axes as dynamic.
in_dformats	dict. Describes input data formats.
yolo_decode_include	bool. Runs YOLO decode on NPU when `True`.
kwargs	dict. Additional compiler arguments.

Definition at line 33 of file frontend.py.

qubee.frontend.Model_Dict.compile	(		self,
		str	calib_data_path = None,
		str	save_path = None,
		int	save_subgraph_type = 0,
			output_subgraph_path = "",
		str	model_nickname = None,
			save_sample = False,
			use_random_calib = False,
			inference_scheme = "single",
			cpu_offload = False,
			singlecore_compile = False,
			optimize_option = 0,
		str	sample_dtype = "float",
			preprocess_dict = None,
			buffer_mode = 1,
			quant_config = QuantConfig.default_config(),
		**	kwargs )

Compile the wrapped model into an MXQ artifact.

Parameters

calib_data_path	string. Calibration dataset directory or meta file (.txt/.json) listing NumPy samples.
save_path	string. Output MXQ filename.
save_subgraph_type	int. Controls optional MBLT exports: 0 disables; 1 saves graph structure; 2 saves structure and weights; 3 splits structure into multiple subgraphs; 4 splits structure and weights.
output_subgraph_path	string. Destination for exported .mblt when `save_subgraph_type` is 1–4.
model_nickname	string. Friendly name stored in qubee for faster recompile.
save_sample	bool. Persists inputs/outputs for debugging in ./sampleInOut.
use_random_calib	bool. Uses randomly generated calibration samples when `True`.
inference_scheme	string. NPU inference configuration ("single", "multi", "global", etc.).
cpu_offload	bool. Enables CPU offload for unsupported groups.
singlecore_compile	bool. Forces single-core compilation for large DRAM usage.
optimize_option	int. Compiler optimization selector.
sample_dtype	string. Data type used for saved sample outputs.
preprocess_dict	dict. Additional preprocessing metadata.
buffer_mode	int. Buffer serialization mode (0 naive, 1 mmap).
quant_config	QuantConfig. Quantization configuration object.
kwargs	dict. Additional arguments forwarded to the compiler.

Definition at line 67 of file frontend.py.

The documentation for this class was generated from the following file: