Quantization-aware training conv1D, LSTM support

It seems this may partially answer my question.