Nov 9 4 days ago

1. 14:11

   Upgrade: ViuAI-UltraDeep 32 Layers (8.96M Total / 4.25M Active / 52.6% Sparsity)

   ViuAI
   Verified signature

   96a059a
2. 13:59

   NextGen: DeepSeek-V4 MLA + Kimi Fine-Grained MoE + mHC + Muon Optimizer (1.58M Total / 0.98M Active)

   ViuAI
   Verified signature

   9e7ff42
3. 13:50

   Upgrade: GQA + QK-Norm + FlashAttention + Deep & Narrow (6 layers, 1.11M params)

   ViuAI
   Verified signature

   7324bad
4. 13:31

   Initial upload: ViuAI-1M architecture (1.18M params)

   ViuAI
   Verified signature

   7f1c419