Back to results
US 2026/0219841 A1Published application
Randomized block-level scaling of neural network tensors
AssigneeNVIDIA Corporation
InventorsChao Ni, Mengdi Wang, Paulius Micikevicius, June Yang
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques to scale tensors of one or more neural networks. In at least one embodiment, a processor includes one or more circuits to cause two or more scaling factors of two or more corresponding portions of one or more tensors to be scaled using one or more randomly generated bits.
Claims20
A processor comprising:
- one or more circuits to cause two or more scaling factors of two or more corresponding portions of one or more tensors to be scaled using one or more randomly generated bits.
A method, comprising:
- causing two or more scaling factors of two or more corresponding portions of one or more tensors to be scaled using one or more randomly generated bits.
A system, comprising:
- one or more processors comprising one or more circuits to cause two or more scaling factors of two or more corresponding portions of one or more tensors to be scaled using one or more randomly generated bits;
- and
- one or more memories to store the two or more corresponding portions of the one or more tensors.
Dependencies are inferred from the claim text.