Fine-grain compute communication execution for deep learning frameworks

S Sridharan, D Mudigere - US Patent App. 15/869,502, 2018 - Google Patents
One embodiment provides for a system to configure distributed training of a neural network.
The system includes memory to store a library to facilitate transmission of data during …

Methods and apparatus for deep learning network execution pipeline on multi-processor platform

L Yang, A Yao - US Patent 11,461,105, 2022 - Google Patents
Methods and systems are disclosed using an execution pipeline on a multi-processor
platform for deep learning network execution. In one example, a network workload analyzer …

Intelligent data transmission by network device agent

S Pasupuleti - US Patent 10,536,505, 2020 - Google Patents
In one aspect, a system for intelligent monitoring of a network device in a monitored
environment includes a processor; a memory; and one or more modules stored in the …

Matrix-factorization based gradient compression

M Cho, V Muthusamy - US Patent 11,182,457, 2021 - Google Patents
Matrix factorization based gradient compression may be applied to an allreduce operation to
improve efficiency including the elimination of unnecessary meta data while maintaining …

Dynamic network bandwidth in distributed deep learning training

W Zhang, X Cui, A Kayi, A Buyuktosunoglu - US Patent 11,886,969, 2024 - Google Patents
Embodiments of a method are disclosed. The method includes performing distributed deep
learning training on a batch of training data. The method also includes determining training …

Dynamic computation in decentralized distributed deep learning training

W Zhang, X Cui, A Kayi, A Buyuktosunoglu - US Patent 11,875,256, 2024 - Google Patents
Embodiments of a method are disclosed. The method includes performing decentralized
distributed deep learning training on a batch of training data. Additionally, the method …

Methods and apparatus for deep learning network execution pipeline on multi-processor platform

L Yang, A Yao - US Patent 11,868,782, 2024 - Google Patents
Methods and systems are disclosed using an execution pipeline on a multi-processor
platform for deep learning network execution. In one example, a network workload analyzer …

Systems and methods for error recovery

B Pudipeddi, M Mesmakhosroshahi, J **… - US Patent …, 2023 - Google Patents
2021-12-16 Assigned to MICROSOFT TECHNOLOGY LICENSING, LLC reassignment
MICROSOFT TECHNOLOGY LICENSING, LLC ASSIGNMENT OF ASSIGNORS INTEREST …

Dynamic computation rates for distributed deep learning

W Zhang, X Cui, A Kayi, A Buyuktosunoglu - US Patent 11,977,986, 2024 - Google Patents
Embodiments of a method are disclosed. The method includes performing distributed deep
learning training on multiple batches of training data using corresponding learners …

Systems and methods for error recovery

B Pudipeddi, M Mesmakhosroshahi, J **… - US Patent …, 2022 - Google Patents
2020-03-27 Assigned to MICROSOFT TECHNOLOGY LICENSING, LLC reassignment
MICROSOFT TECHNOLOGY LICENSING, LLC ASSIGNMENT OF ASSIGNORS INTEREST …