minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[diffusion] Reuse SANA fast paths in SANA-Video BCG (#35961)

Browse Source
This commit is contained in:
Xiaoyu Zhang
2026-08-24 08:47:04 +08:00
committed by GitHub
parent 20064623ab
commit f4448e677f
5 changed files with 85 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/docs/sglang-diffusion/api/cli.mdx
+5 -3
View File
@@ -237,9 +237,11 @@ sglang serve --model-path meituan-longcat/LongCat-Image \
--port 30010
```
SANA-Video supports the same path for fixed-resolution serving. Its default
text stage emits a fixed 300-token prompt shape, so the runtime reuses one
graph across prompt lengths without padding it to a generic text bucket:
SANA-Video supports the same path for fixed-shape serving. Its default text
stage emits a fixed 300-token prompt shape, so the runtime reuses one graph
across prompt lengths without padding it to a generic text bucket. BCG also
requires the serving frame count to match warmup; the example captures the
model's default 81-frame signature:
```bash Command
sglang serve \
docs/docs/sglang-diffusion/fused_kernels.mdx
+1 -2
View File
@@ -147,9 +147,8 @@ Kernels are written against a specific eager chain in a specific model, so cover
| LTX-2 | QK-norm + split RoPE, ada-values split, RMSNorm+modulate, modulate, residual-gate add, linear+GELU |
| LTX-2.5 decoder | paired 3D RoPE with shared axis-table cache |
| HunyuanVideo | QKV+RoPE pack, strided QK RMSNorm, linear+GELU |
| SANA-Video | paired fp64 interleaved RoPE |
| Sana | LN+modulate, GLUMB bias+SiLU / bias+GLU, residual-gate add |
| SANA-Video | Packed QKV/KV; BF16-input linear attention at `quality=high` |
| SANA-Video | Packed QKV/KV; paired fp64 interleaved RoPE; LN+modulate, GLUMB bias+SiLU / bias+GLU, and residual-gate add during BCG; BF16-input linear attention at `quality=high` |
| Sana-WM | bidirectional gated delta-net, fused QK inverse-RMS |
| Wan | temb table slices; VAE cat+pad and DupUp3D add, `channels_last_3d` RMSNorm+SiLU |
| Cosmos3 / Krea2 / MiniMax-H3 | QK-norm + RoPE (Krea2 also CuTe-DSL norm+scale/shift; MiniMax-H3 also indexed modulation) |