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

[diffusion] doc: sync CFG and tracing documentation (#39883)

Browse Source 
Co-authored-by: Mick Qian <mickqian@users.noreply.github.com>
This commit is contained in:
Mick
2026-09-18 18:03:40 +08:00
committed by GitHub
co-authored by Mick Qian
parent a6cf05817f
commit 9784d5f979
3 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/cookbook/diffusion/MiniMax/MiniMax-H3.mdx
+1 -1
View File
@@ -1342,7 +1342,7 @@ fold decision is not node-boundary aware:
- V2V uses `task: "ref2va"` with a `video` or `video_audio` reference; it is served by the `Ref2VA` partition and is not a separate public task value.
- `conditions[].start_time_seconds` selects a non-negative offset for a video reference. Its visual and audio streams are always sought together.
- Ref2VA condition order is semantic and must match the one-based material tags in the prompt. For Ref2VA, `target.aspect_ratio: "auto"` resolves to the model's 16:9 fallback rather than inheriting a reference asset's geometry.
- The distilled pipeline uses a single denoising branch, so CFG parallelism does not apply. Do not enable it: `--enable-cfg-parallel true` or `--cfg-parallel-size` greater than 1 is rejected instead of duplicating the positive branch. Explicitly disabling CFG, or setting its size to 1, remains a valid no-op.
- The distilled pipeline uses a single denoising branch, so CFG parallelism does not apply. Keep it disabled: a CFG degree greater than 1 is accepted for request compatibility, but the extra ranks only recompute the same branch and do not accelerate the request. Explicitly disabling CFG, or setting its size to 1, remains a valid no-op.
- The released visual VAE quality recipe uses overlapping tiled decode. SGLang keeps that recipe by default and distributes complete tiles across the decode group; this changes scheduling, not the computation inside each tile.
- H3 rejects `--vae-config.parallel-decode-mode spatial` and `spatial_shard`: validation found output mismatches. Use the default released tiled recipe.
- Keep the default `--encoder-parallel auto`. With the server’s default `batching_max_size` of 1, single-node H100/H200/B200/B300 recipes with peer-to-peer access fold the Qwen text encoder over otherwise idle Ulysses ranks. This is separate from DiT tensor parallelism. A pure-TP recipe already shards the encoder over its TP group and does not add a world fold.
docs/cookbook/diffusion/SANA-WM/SANA-WM.mdx
+1 -1
View File
@@ -192,7 +192,7 @@ Notes on launch behavior:
- **Default endpoint** is `127.0.0.1:30000` (`--host` / `--port` override).
- **CPU offload flags are optional.** `--dit-cpu-offload`, `--text-encoder-cpu-offload`, and `--image-encoder-cpu-offload` are available; defaults are auto-adjusted from GPU memory (GPUs under 30 GB get more aggressive offloading).
- **Multi-GPU realtime.** Prefer explicit sequence parallelism (`--sp-degree` equal to the number of GPUs for a single session). Do not enable CFG parallel for the realtime profile: the default realtime request uses `guidance_scale=1.0`, while CFG parallel requires active cond/uncond branches.
- **Multi-GPU realtime.** Prefer explicit sequence parallelism (`--sp-degree` equal to the number of GPUs for a single session). Do not enable CFG parallel for the realtime profile: the default request uses `guidance_scale=1.0`, so the extra CFG ranks only recompute the same branch.
- **FSDP.** Use `--use-fsdp-inference` only when you specifically need weight sharding for memory. For the low-latency realtime profile, prefer keeping components resident and using SP first.
- **Warmup.** Server warmup is **automatically skipped** for the realtime pipeline — a synthetic warmup request has no WebSocket session, so the server detects the registered realtime adapter and skips it. No explicit `--warmup-mode` setting is needed.