Skip to main content
GPUBeat › Archive

/Tag: quantization

Advancements in on-device AI capabilities — Vikas Chandra, Meta
Frontier Models 1d

Meta’s Vision for Intelligent Personal Devices Takes Center Stage

At the Embedded Vision Summit, Meta's Vikas Chandra highlighted efforts to bring advanced AI to personal devices, prioritizing context over model size.

GD
GPUBeat DeskMay 193 min
More from this archive
Frontier Models 2d

NVIDIA Breaks New Ground With 4-Bit LLM Pretraining Methodology

NVIDIA's new NVFP4 methodology for training large language models at 4-bit precision achieves near-FP8 performance, setting a new…