Patent Leaderboard
USPTO Patent Rankings Data through Dec 31, 2025
Home › Lossless exponent and lossy mantissa weight compression for training deep neural networks

Lossless exponent and lossy mantissa weight compression for training deep neural networks

US Patent 11615301 · Granted Mar 28, 2023

Estimated economic value: $376,835,000 ⓘ

Assignee

Inventors

View full patent text on Google Patents →

More Patents by These Inventors

Stash balancing in model parallelism2025
Token packing for sequence models2025
Forcing weights of transformer model layers2025
Hardware-assisted gradient optimization using streamed gradients2025
Hardware-assisted gradient optimization using streamed gradients2025
Hardware-assisted gradient optimization using streamed gradients2025
Systems and methods for hardware acceleration of data masking2025
Compressing and decompressing data for language models2024
Systems and methods for managing undesired effects produced by horticultural lig2024
Systems and methods for monitoring and managing reactive power from horticultura2024
Method and system for simulating and implementing horticultural lighting scenari2024
Determining position values for transformer models2024