`neural_compressor.adaptor.tf_utils.graph_rewriter.int8.post_quantized_op_cse`

Post CSE Graph Rewriter.

Module Contents

Classes

PostCseOptimizer

Remove duplicated nodes like shared quantizev2 and const to decrease the output model size.

class neural_compressor.adaptor.tf_utils.graph_rewriter.int8.post_quantized_op_cse.PostCseOptimizer(model)[source]: Remove duplicated nodes like shared quantizev2 and const to decrease the output model size.