5 repos
TsinghuaC3I/ZEDA
Post-Trained MoE Can Skip Half Experts via Self-Distillation
40
17 commits
TsinghuaC3I/ZEDA-Qwen3-30B-A3B-Dynamic
No description
1
4 commits
TsinghuaC3I/ZEDA-GLM-4.7-Flash-Dynamic
2
10 commits
zenlm/zen5
zen5 MoDE — Mixture of Diverse Experts. 3.1T params, 0.8-100B+ active, complexity-aware routing…
0
2 commits
