From aca0d01d3f55ec7dd46d68d1c6d01ae7900b8021 Mon Sep 17 00:00:00 2001 From: Dongjie Zou <85092850+baonudesifeizhai@users.noreply.github.com> Date: Wed, 3 Dec 2025 10:13:25 -0500 Subject: [PATCH] [diffusion] doc: add vae path to cli doc#14004 (#14355) Co-authored-by: BBuf <1182563586@qq.com> Co-authored-by: Mick --- python/sglang/multimodal_gen/docs/cli.md | 1 + 1 file changed, 1 insertion(+) diff --git a/python/sglang/multimodal_gen/docs/cli.md b/python/sglang/multimodal_gen/docs/cli.md index 4b0f29c72..1c054e466 100644 --- a/python/sglang/multimodal_gen/docs/cli.md +++ b/python/sglang/multimodal_gen/docs/cli.md @@ -13,6 +13,7 @@ The SGLang-diffusion CLI provides a quick way to access the inference pipeline f ### Server Arguments - `--model-path {MODEL_PATH}`: Path to the model or model ID +- `--vae-path {VAE_PATH}`: Path to a custom VAE model or HuggingFace model ID (e.g., `fal/FLUX.2-Tiny-AutoEncoder`). If not specified, the VAE will be loaded from the main model path. - `--num-gpus {NUM_GPUS}`: Number of GPUs to use - `--tp-size {TP_SIZE}`: Tensor parallelism size (only for the encoder; should not be larger than 1 if text encoder offload is enabled, as layer-wise offload plus prefetch is faster) - `--sp-size {SP_SIZE}`: Sequence parallelism size (typically should match the number of GPUs)