[Diffusion] Support peak memory record in offline generate and serving (#15610)

2025-12-22 21:21:21 +08:00
parent 828dec1c7c
commit d77f3fccbf
8 changed files with 88 additions and 15 deletions
--- a/python/sglang/multimodal_gen/runtime/entrypoints/openai/protocol.py
+++ b/python/sglang/multimodal_gen/runtime/entrypoints/openai/protocol.py
@@ -14,6 +14,7 @@ class ImageResponseData(BaseModel):
 class ImageResponse(BaseModel):
    created: int = Field(default_factory=lambda: int(time.time()))
    data: List[ImageResponseData]
+    peak_memory_mb: Optional[float] = None


 class ImageGenerationsRequest(BaseModel):
@@ -50,6 +51,7 @@ class VideoResponse(BaseModel):
    completed_at: Optional[int] = None
    expires_at: Optional[int] = None
    error: Optional[Dict[str, Any]] = None
+    peak_memory_mb: Optional[float] = None


 class VideoGenerationsRequest(BaseModel):