most citedTrained Quantization Thresholds for Accurate and Efficient Fixed-Point Inference of Deep Neural Networks

55 citations · 55 across the 1 of their papers we have counts for

collaborators

1 paper

cs.CV2019★ 55 cited

Trained Quantization Thresholds for Accurate and Efficient Fixed-Point Inference of Deep Neural Networks

Sambhav R. Jain, Albert Gural, Michael Wu +1

We propose a method of training quantization thresholds (TQT) for uniform symmetric quantizers using standard backpropagation and gradient descent. Contrary to prior work, we show…