ViuAI
/

ViuMini-Dense-360M

Text Generation

Model card Files Files and versions

ViuMini-Dense-360M / model /scripts

168 kB

Ctrl+K

Ctrl+K

2 contributors

History: 70 commits

ViuAI's picture

fix: restore grad_ckpt and safe micro 8 accum 8 for zero-OOM guarantee on T4

96be75f verified about 7 hours ago

model.py

217 Bytes
refactor: consolidate pure 32L Dense architecture, purge MoE files, audit hardening about 8 hours ago
push_to_hf.py

2.68 kB
refactor: consolidate pure 32L Dense architecture, purge MoE files, audit hardening about 8 hours ago
toy_router_test.py

5.62 kB
step3: add F(0.10/50) + G(0.05/100) 17 days ago
train.py

101 kB
fix: restore grad_ckpt and safe micro 8 accum 8 for zero-OOM guarantee on T4 about 7 hours ago
verify_1_5b_arch.py

2.97 kB
Add 42-layer Viu-1.5B-MoE verification test suite 9 days ago
viu1_dense.py

12.7 kB
fix: restore grad_ckpt and safe micro 8 accum 8 for zero-OOM guarantee on T4 about 7 hours ago
viu1_moe.py

17.8 kB
P0 fix: gradient checkpointing, chunked cross-entropy, and vectorized MoE dispatch 1 day ago
viu_moe.py

24.5 kB
Vectorize MoE dispatch (argsort+bincount): eliminate 2,688 CUDA kernel launches per pass 2 days ago