Company Filing History:
Years Active: 2021-2025
Title: Slawomir Kierat: Innovator in Tensor Processing and Data Conversion
Introduction
Slawomir Kierat is a notable inventor based in Mountain View, CA, who has made significant contributions to the field of artificial intelligence and neural networks. With a focus on improving data processing techniques, he holds two patents that showcase his innovative approach to tensor processing and data conversion.
Latest Patents
Kierat's latest patents include groundbreaking work on "Tensor processing using low precision format." This invention addresses the challenges of training artificial neural networks by utilizing a reduced precision data format, such as float16. The technique involves rescaling tensor values before matrix operations to prevent overflow and underflow, ensuring accuracy throughout the computations. Another significant patent is "Automated methods for conversions to a lower precision data format." This invention enhances data compression and conversion techniques, improving the inferencing capabilities of artificial neural networks using reduced precision formats like INT8. By generating candidate conversions and employing a quality measure, this method identifies the most accurate conversion for effective computations.
Career Highlights
Slawomir Kierat is currently employed at Nvidia Corporation, a leading company in graphics processing and artificial intelligence technologies. His work at Nvidia has allowed him to push the boundaries of what is possible in the realm of neural networks and data processing.
Collaborations
Kierat collaborates with talented individuals such as Alex Fit-Florea and Boris Ginsburg, contributing to a dynamic team focused on advancing technology in artificial intelligence.
Conclusion
Slawomir Kierat's innovative patents and contributions to the field of artificial intelligence highlight his role as a key inventor in tensor processing and data conversion. His work continues to influence advancements in neural network training and data processing techniques.