PTQ后转TVM失败

[Xrigok]

PTQ后转TVM失败： 在追溯的时候打印MQBench-main/mqbench/deploy/common.py里的self.graph.initializer，发现不包括对输入的伪量化节点，导致self.onnx_model.graph.node里对输入的伪量化节点在self.onnx_model.initializer里找不到。 请问下这个问题怎么解决，在qat转TVM里是包括对输入的伪量化节点，但ptq却没有，会是导出onnx的问题吗？ 底下是进行PTQ时对graph.initializer的name的打印，可以发现不包括对输入的伪量化节点： conv1.weight conv1.bias conv1.weight_fake_quant.scale conv1.weight_fake_quant.zero_point layer1.0.conv1.weight layer1.0.conv1.bias layer1.0.conv1.weight_fake_quant.scale layer1.0.conv1.weight_fake_quant.zero_point layer1.0.conv2.weight layer1.0.conv2.bias layer1.0.conv2.weight_fake_quant.scale layer1.0.conv2.weight_fake_quant.zero_point layer1.1.conv1.weight layer1.1.conv1.bias layer1.1.conv1.weight_fake_quant.scale layer1.1.conv1.weight_fake_quant.zero_point layer1.1.conv2.weight layer1.1.conv2.bias layer1.1.conv2.weight_fake_quant.scale layer1.1.conv2.weight_fake_quant.zero_point layer2.0.conv1.weight layer2.0.conv1.bias layer2.0.conv1.weight_fake_quant.scale layer2.0.conv1.weight_fake_quant.zero_point layer2.0.conv2.weight layer2.0.conv2.bias layer2.0.conv2.weight_fake_quant.scale layer2.0.conv2.weight_fake_quant.zero_point layer2.0.downsample.0.weight layer2.0.downsample.0.bias layer2.0.downsample.0.weight_fake_quant.scale layer2.0.downsample.0.weight_fake_quant.zero_point layer2.1.conv1.weight layer2.1.conv1.bias layer2.1.conv1.weight_fake_quant.scale layer2.1.conv1.weight_fake_quant.zero_point layer2.1.conv2.weight layer2.1.conv2.bias layer2.1.conv2.weight_fake_quant.scale layer2.1.conv2.weight_fake_quant.zero_point layer3.0.conv1.weight layer3.0.conv1.bias layer3.0.conv1.weight_fake_quant.scale layer3.0.conv1.weight_fake_quant.zero_point layer3.0.conv2.weight layer3.0.conv2.bias layer3.0.conv2.weight_fake_quant.scale layer3.0.conv2.weight_fake_quant.zero_point layer3.0.downsample.0.weight layer3.0.downsample.0.bias layer3.0.downsample.0.weight_fake_quant.scale layer3.0.downsample.0.weight_fake_quant.zero_point layer3.1.conv1.weight layer3.1.conv1.bias layer3.1.conv1.weight_fake_quant.scale layer3.1.conv1.weight_fake_quant.zero_point layer3.1.conv2.weight layer3.1.conv2.bias layer3.1.conv2.weight_fake_quant.scale layer3.1.conv2.weight_fake_quant.zero_point layer4.0.conv1.weight layer4.0.conv1.bias layer4.0.conv1.weight_fake_quant.scale layer4.0.conv1.weight_fake_quant.zero_point layer4.0.conv2.weight layer4.0.conv2.bias layer4.0.conv2.weight_fake_quant.scale layer4.0.conv2.weight_fake_quant.zero_point layer4.0.downsample.0.weight layer4.0.downsample.0.bias layer4.0.downsample.0.weight_fake_quant.scale layer4.0.downsample.0.weight_fake_quant.zero_point layer4.1.conv1.weight layer4.1.conv1.bias layer4.1.conv1.weight_fake_quant.scale layer4.1.conv1.weight_fake_quant.zero_point layer4.1.conv2.weight layer4.1.conv2.bias layer4.1.conv2.weight_fake_quant.scale layer4.1.conv2.weight_fake_quant.zero_point fc.weight fc.bias 691

[github-actions[bot]]

This issue has not received any updates in 120 days. Please reply to this issue if this still unresolved!