RWKV-LM
RWKV-LM copied to clipboard
Where's the cuda backward function for v7?
trafficstars
As stated in the title, seems cuda_backward and kernel_backward is missing in RWKV-v7/cuda/wkv7.cu and RWKV-v7/cuda/wkv7_op.cpp
If you try to train a model with RWKV-v7 I'm getting:
NotImplementedError: You must implement either the backward or vjp method for your custom autograd.Function to use it with backward mode AD.