From f3741ac2b94cb43356e9b7b1c193feb8568df52d Mon Sep 17 00:00:00 2001 From: JohnJan Date: Sat, 10 Jan 2026 19:02:32 +0800 Subject: [PATCH] [Doc]: update vllm serve param and base64 data truncation (#718) Signed-off-by: wuzhongjian Signed-off-by: Hongsheng Liu Co-authored-by: Hongsheng Liu Co-authored-by: WeiQing Chen <40507679+david6666666@users.noreply.github.com> --- docs/user_guide/examples/online_serving/text_to_image.md | 6 ++++-- examples/online_serving/text_to_image/README.md | 6 ++++-- 2 files changed, 8 insertions(+), 4 deletions(-) diff --git a/docs/user_guide/examples/online_serving/text_to_image.md b/docs/user_guide/examples/online_serving/text_to_image.md index adc9c5c30..877f20a4c 100644 --- a/docs/user_guide/examples/online_serving/text_to_image.md +++ b/docs/user_guide/examples/online_serving/text_to_image.md @@ -12,6 +12,8 @@ This example demonstrates how to deploy Qwen-Image model for online image genera ```bash vllm serve Qwen/Qwen-Image --omni --port 8091 ``` +!!! note + If you encounter Out-of-Memory (OOM) issues or have limited GPU memory, you can enable VAE slicing and tiling to reduce memory usage, --vae-use-slicing --vae-use-tiling ### Start with Parameters @@ -43,7 +45,7 @@ curl -s http://localhost:8091/v1/chat/completions \ "true_cfg_scale": 4.0, "seed": 42 } - }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2 | base64 -d > output.png + }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > output.png ``` ### Method 2: Using Python Client @@ -146,7 +148,7 @@ Use `extra_body` to pass generation parameters: ```bash # Extract base64 from response and decode to image -cat response.json | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2 | base64 -d > output.png +cat response.json | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > output.png ``` ## File Description diff --git a/examples/online_serving/text_to_image/README.md b/examples/online_serving/text_to_image/README.md index 6f835e66e..a4f1ad633 100644 --- a/examples/online_serving/text_to_image/README.md +++ b/examples/online_serving/text_to_image/README.md @@ -9,6 +9,8 @@ This example demonstrates how to deploy Qwen-Image model for online image genera ```bash vllm serve Qwen/Qwen-Image --omni --port 8091 ``` +!!! note + If you encounter Out-of-Memory (OOM) issues or have limited GPU memory, you can enable VAE slicing and tiling to reduce memory usage, --vae-use-slicing --vae-use-tiling ### Start with Parameters @@ -40,7 +42,7 @@ curl -s http://localhost:8091/v1/chat/completions \ "true_cfg_scale": 4.0, "seed": 42 } - }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2 | base64 -d > output.png + }' | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > output.png ``` ### Method 2: Using Python Client @@ -143,7 +145,7 @@ Use `extra_body` to pass generation parameters: ```bash # Extract base64 from response and decode to image -cat response.json | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2 | base64 -d > output.png +cat response.json | jq -r '.choices[0].message.content[0].image_url.url' | cut -d',' -f2- | base64 -d > output.png ``` ## File Description