Show HN: Maple MoE Pruning for Gemma models
A project demonstrating sub-2-BPW Mixture-of-Experts pruning to run the Gemma 4 26B model at high speeds on a 6GB GPU.
A project demonstrating sub-2-BPW Mixture-of-Experts pruning to run the Gemma 4 26B model at high speeds on a 6GB GPU.