Haste: a fast, simple, and open RNN library
LayerNormGRU
).state
argument to PyTorch layers to specify initial state.get_weights
method to extract weights from RNN layers (TensorFlow).to_native_weights
and from_native_weights
to PyTorch API for LSTM
and GRU
layers.h
must not be transposed before passing it to gru::BackwardPass::Iterate
.cublasHandle_t
.This release focuses on LSTM performance.
lstm::ForwardPass::Run
, lstm::BackwardPass::Run
).h
must not be transposed before passing it to lstm::BackwardPass::Iterate
.dv
does not need to be allocated and v
must be passed instead to lstm::BackwardPass::Iterate
.Initial release.