Second order derivates for lstm_cell on GPU

I’m trying to compute the second-order derivative for an LSTMCell. If I do it on CPU, everything is fine. But, if I do it on GPU, I get the following error:
RuntimeError: derivative for _thnn_fused_lstm_cell_backward is not implemented

Is there any solution to compute it on GPU?

PyTorch version: Commit ae1a972 (installed from source)

1 Like

Did you solve this problem??