search

raywan-110 / Quantization-DNNs

simulate Tensorflow Lite with Pytorch API
MIT License
1 stars 0 forks source link

Running QAT.py gives error Tensor Descriptor only supports double, float and half tensors #1

Open ChidanandKumarKS opened 2 years ago

ChidanandKumarKS commented 2 years ago

@raywan-110 Kindly help Running QAT.py gives error Tensor Descriptor only supports double, float and half tensors which is thrown during inference quantize_inference(model,test_iter)

ralohcslive commented 1 year ago

Hi, I've got a different error, " File C:\Quantization-DNNs\Qmodule.py:413 in freeze self.conv_module.weight.data = self.qw.quantize_tensor(weight.data.requiresgrad().long())

RuntimeError: data set to a tensor that requires gradients must be floating point or complex dtype "

How did you fix this ?