Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers Paper • 2605.18106 • Published 3 days ago • 1
Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers Paper • 2605.18106 • Published 3 days ago • 1
METAGENE-1: Metagenomic Foundation Model for Pandemic Monitoring Paper • 2501.02045 • Published Jan 3, 2025 • 22