micronet, a model compression and deploy lib. compression: 1、quantization: quantization-aware-training(QAT), High-Bit(>2b)(DoReFa/Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference)、Low-Bit(≤2b)/Ternary and Binary(TWN/BNN/XNOR-Net); post-training-quantization(PTQ), 8-bit(tensorrt); 2、 pruning: normal、regular and group convolutional channel pruning; 3、 group convolution structure; 4、batch-normalization fuse for quantization. deploy: tensorrt, fp32/fp16/int8(ptq-calibration)、op-adapt(upsample)、dynamic_shape
EdwardVincentMa
commented
3 years ago
你好，我的选项是bnfused，在原始nin_gc中，例如第三个卷积，里面有channel_shuffle and shuffle_groups选项，但是，当卷积替换为QuantbBNFusedConv2d的时候缺没有这个选项了，里面并没有reshape+trans+reshape，这是为什么呢？