quark.torch.quantization.graph.processor.insert_quantizer

Module Contents

Functions

quark.torch.quantization.graph.processor.insert_quantizer.insert_quantizer(model: torch.fx.GraphModule) → torch.fx.GraphModule

Inserts FakeQuantize call_module nodes in the graph for input and/or output quantization, if necessary, based on the quantization_annotation metadata attached to nodes.