体验新版 GitCode，发现更多精彩内容 >>
* Add a fake_quantize_op, which quantize an input tensor to a tensor with lower bits.
