Don't have Telegram yet? Try it now!
https://hgpu.org/?p=17285
MobiRNN: Efficient Recurrent Neural Network Execution on Mobile GPU