chop.passes.transform.quantize#
quantize_transform_pass#
- chop.passes.graph.transforms.quantize.quantize_transform_pass(graph, pass_args=None)[source]#
Apply quantization transformation to the given graph.
- Parameters:
graph (MaseGraph) – The input graph to be transformed.
pass_args (dict, optional) – Additional arguments for the transformation.
- Returns:
The transformed graph.
- Return type:
tuple
- Raises:
ValueError – If the quantize “by” argument is unsupported.
- pass_args
by -> str : different quantization schemes choose from [“type”, “name”, “regx_name”]
summarize_quantization_analysis_pass#
- chop.passes.graph.transforms.quantize.summarize_quantization_analysis_pass(graph, pass_args={'original_graph': None, 'save_dir': None}) None[source]#
Summarizes the quantization analysis pass. Apply quantization transformation to the given graph.
- Parameters:
graph (MaseGraph) – The input graph to be transformed.
pass_args (dict, optional) – Additional arguments for the transformation.
- Returns:
The transformed MaseGraph.
- Return type:
tuple