Pruning giant AI models beats training small ones from scratch

Original: “SlimQwen: Exploring the Pruning and Distillation in Large MoE Model Pre-training”

Source: x.com ↗