The rapid advancement of large language and multimodal models has made computing resource and energy efficiency key bottlenecks in real-world intelligent systems. Low bit-width numerical representations provide a promising approach to substantially reduce computational load, memory footprint, and energy consumption.
As one of the IEEE ICME 2026 Grand Challenges, this challenge is organized by the Global Computing Consortium (GCC) and aims to promote algorithmic and system-level innovations in low bit-width training and inference for large language models.
The challenge cordially invites AI algorithm researchers, application and system developers, and technology practitioners and visionaries dedicated to computational efficiency from around the world to participant. Together, we seek to bridge the gap between theoretical quantization methods and real-world deployment constraints, and to develop innovative, reproducible, and engineering-ready solutions for efficient large model computation.
For detailed information and registration, please visit:https://challenge.gccorg.com/