Title:
DATA PROCESSING METHOD AND DEVICE FOR NEURAL NETWORK MODEL, APPARATUS, AND STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2021/164506
Kind Code:
A1
Abstract:
Disclosed in embodiments of the present invention are a data processing method and device for a neural network model, an apparatus, and a storage medium. The method comprises: acquiring multiple neural network operators in a neural network model; performing fusion on the multiple neural network operators according to a preset rule, and obtaining fused neural network operators; combining the fused neural network operators into a calculation instruction; and using a calculation engine to perform calculation with respect to the calculation instruction.
Inventors:
HUANG JIONGKAI (CN)
TSOI KUEN HUNG (CN)
NIU XINYU (CN)
TSOI KUEN HUNG (CN)
NIU XINYU (CN)
Application Number:
PCT/CN2021/073758
Publication Date:
August 26, 2021
Filing Date:
January 26, 2021
Export Citation:
Assignee:
SHENZHEN CORERAIN TECH CO LTD (CN)
International Classes:
G06N3/02
Foreign References:
CN111260019A | 2020-06-09 | |||
CN109754073A | 2019-05-14 | |||
CN110689116A | 2020-01-14 | |||
CN110321999A | 2019-10-11 |
Attorney, Agent or Firm:
BEYOND ATTORNEYS AT LAW (CN)
Download PDF:
Previous Patent: ELECTRONIC DEVICE
Next Patent: SCHEDULING METHOD, SCHEDULING ALGORITHM TRAINING METHOD AND RELATED SYSTEM, AND STORAGE MEDIUM
Next Patent: SCHEDULING METHOD, SCHEDULING ALGORITHM TRAINING METHOD AND RELATED SYSTEM, AND STORAGE MEDIUM