RFC-0019 Extending PyTorch Quantization to Custom Backends

Hi, just wondering, has RFC-0019 been accepted? What described in the documentation seems really helpful for implementing a custom backend for PyTorch Quantization. So kindly wanted to know if there is an issue/a page that tracks the progress of this RFC.

Hi @April_Hu I believe that this work is currently in progress, but Iā€™m not aware of a tracker. Jerry might have more information.

yeah this is accepted and implemented in the fx graph mode quantization flow (convert_to_reference_fx), but we have a new flow coming up, so if you want start integration it might be better to go with the new flow (that also implements this design). Here are some docs for the new flow: