Title:
加速された量子化積和演算
Document Type and Number:
Japanese Patent JP6946572
Kind Code:
B2
Abstract:
Disclosed herein are techniques for accelerating convolution operations or other matrix multiplications in applications such as neural network. In one example, an apparatus comprises a first circuit, a second circuit, and a third circuit. The first circuit is configured to: receive first values in a first format, the first values being generated from one or more asymmetric quantization operations of second values in a second format, and generate difference values based on subtracting a third value from each of the first values, the third value representing a zero value in the first format. The second circuit is configured to generate a sum of products in the first format using the difference values. The third circuit is configured to convert the sum of products from the first format to the second format based on scaling the sum of products with a scaling factor.
Inventors:
Vantries, Dana Michel
Fan, Randy
Diamant, Ron
Elmer, Thomas
Amilineni, Sundeep
Fan, Randy
Diamant, Ron
Elmer, Thomas
Amilineni, Sundeep
Application Number:
JP2020551488A
Publication Date:
October 06, 2021
Filing Date:
March 20, 2019
Export Citation:
Assignee:
Amazon Technologies, Inc.
International Classes:
G06N10/00; G06N3/02
Domestic Patent References:
| JP2018010618A | ||||
| JP6293963B1 |
Other References:
JACOB, Benoit et al.,Quantization and training of neural networks for efficient integer-arithmetic-only inference [online],2017年12月15日,[検索日 2021.03.31], Internet:
Attorney, Agent or Firm:
Noriaki Okabe
Jin Aiba
Tetsuya Yaguchi
Jin Aiba
Tetsuya Yaguchi