Chia-Tse Chang
Chia-Tse Chang
Hi, Tianzhi0549: Thanks for the reply. It seems that it can compute targets level by level with the same policy for levels. I was just not sure that some operations...
HI, I think that if you can set the targets corresponding to the right places, it should be fine in both image-first and level-first. But, the field size is matched...
Hi, I guess that you are using PyTorch >= 1.5.0. It seems to be the problem of PyTorch API. I did not meet this in version 1.4.0., and floor_divide seems...
Hi, what is the batch size you set? I remembered that it also happened if the batch size is small when I am doing the experiments. I guess that it...
Hi Peony, sorry that I did not test on CIFAR10. Did you test that training without quantization? And what about the initial learning rate?
Hi, xieydd, 1. yes, you can, it should just change the bit range. 2. I only evaluate the fake-quantization. The evaluation can directly use --evaluate though it runs on the...
Hi, did you pull the submodule from PyTransformer?
Hi, z13974509906: The purpose we build this repository is for general torch.nn layers. So it rarely required multiple inputs. Could you explain the detail or provide some examples for testing...
Hmm, I did not consider the non-tensor output. If you are in a hurry to use the visualization/summary part, you may try some rules to change the rule of the...