ROCm / MIOpen

AMD's Machine Intelligence Library
https://rocm.docs.amd.com/projects/MIOpen/en/latest/
Other
1.09k stars 231 forks source link

[RNN] LSTM backward weight MS #3241

Closed shurale-nkn closed 2 months ago

shurale-nkn commented 2 months ago

Add LSTM backward weight multi-stream solution. Expected small perf improvement for backward weights fp16 - 5%. 2% for full train time