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

Fuse QuantizedMatMul with Requantize/Dequantize Graph Rewriter.

Module Contents

`FuseMatMulRequantizeDequantizeTransformer`	Fuse QuantizedMatMul + Requantize + Dequantize into QuantizedMatMulWithBiasAndDequantize.
`FuseMatMulRequantizeTransformer`	Fuse Quantized MatMul Op with the successor Requantize Op.
`FuseMatMulRequantizeDequantizeNewAPITransformer`	Fuse _QuantizedMatMul + Requantize + Dequantize into _QuantizedMatMul.
`FuseMatMulRequantizeNewAPITransformer`	Fuse newAPI Quantized MatMul Op with the successor Requantize Op.

class neural_compressor.adaptor.tf_utils.graph_rewriter.int8.fuse_matmul_requantize.FuseMatMulRequantizeDequantizeTransformer(model, device='cpu')[source]: Fuse QuantizedMatMul + Requantize + Dequantize into QuantizedMatMulWithBiasAndDequantize.

class neural_compressor.adaptor.tf_utils.graph_rewriter.int8.fuse_matmul_requantize.FuseMatMulRequantizeTransformer(model, device='cpu')[source]: Fuse Quantized MatMul Op with the successor Requantize Op.

class neural_compressor.adaptor.tf_utils.graph_rewriter.int8.fuse_matmul_requantize.FuseMatMulRequantizeDequantizeNewAPITransformer(model, device='cpu')[source]: Fuse _QuantizedMatMul + Requantize + Dequantize into _QuantizedMatMul.

class neural_compressor.adaptor.tf_utils.graph_rewriter.int8.fuse_matmul_requantize.FuseMatMulRequantizeNewAPITransformer(model, device='cpu')[source]: Fuse newAPI Quantized MatMul Op with the successor Requantize Op.