Fusion Support

Currently, this tool mainly implements the following forms of BN fusion (single-operators involved in the following fusion forms must meet the restrictions of Quantization):

• Conv+BN+Scale fusion: Before AMCT-based quantization, the "Convolution+BatchNorm+Scale" composite in the model is fused into "Conv+BN+Scale." The BatchNorm and Scale layers are removed.
• Deconv+BN+Scale fusion: Before AMCT-based quantization, the "Deconvolution+BatchNorm+Scale" composite in the model is fused into "Deconv+BN+Scale." The BatchNorm and Scale layers are removed.
• BN+Scale+Conv fusion: Supported only after PTQ. Before quantization, BN+Scale+Conv fusion is performed on the BatchNorm+Scale+Conv structure in the model. The BatchNorm and Scale layers after fusion are deleted.
• FC+BN+Scale fusion: This function is supported only after quantization. Before quantization, FC+BN+Scale fusion is performed on the InnerProduct+BatchNorm+Scale structure in the model. After the fusion, the BatchNorm and Scale layers are deleted.