intel / nn-hal

44 stars 51 forks source link

Add support for QuantizedLSTM operation #130

Open anikulk opened 3 years ago

anikulk commented 3 years ago

This change also adds support for dequantization for GNA as GNAPlugin does not support Convert operation.

This Q_LSTM operation does not pass the Q_LSTM VTS since we are seeing issues with accuracy in Quantized ops. This is to be fixed