ConvTranspose2d in Adaround
Closed this issue · 4 comments
jmliu206 commented
When I use Adaround to quantize my model, it seems that it is useless to the convtransposed2d.
I check advanced_ptq.py, there is no support for convtransposed2d.
_ADAROUND_SUPPORT_TYPE = (torch.nn.Conv2d, torch.nn.Linear)
_FUSED_TYPE = (nniqat.ConvBnReLU2d, nniqat.ConvBn2d, qnniqat.ConvFreezebn2d, qnniqat.ConvFreezebnReLU2d)
_WEIGHTS_MODULE_TYPE = (torch.nn.Conv2d, torch.nn.Linear)
Could I know how to utilize adaround for convtransposed2d?
Tracin commented
Update _ADAROUND_SUPPORT_TYPE and add convtransposed2d in it.
See if any further error occurs...
github-actions commented
This issue has not received any updates in 120 days. Please reply to this issue if this still unresolved!
jmliu206 commented
Thanks, it works
github-actions commented
This issue has not received any updates in 120 days. Please reply to this issue if this still unresolved!