1

Dockerfile

@@ -0,0 +1,37 @@
+FROM ghcr.io/astral-sh/uv:latest AS uv_bin
+FROM nvidia/cuda:12.4.1-base-ubuntu22.04
+
+# 1. 基础环境配置
+ENV UV_LINK_MODE=copy \
+    UV_COMPILE_BYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    UV_PROJECT_ENVIRONMENT=/app/.venv
+
+COPY --from=uv_bin /uv /uvx /bin/
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+        wget \
+        libcurl4-openssl-dev \
+        build-essential \
+        libgl1 \
+        libglib2.0-0 \
+        ca-certificates \
+        git \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /app
+
+COPY pyproject.toml uv.lock ./
+
+ENV UV_COMPILE_BYTECODE=0
+
+RUN uv sync --frozen --no-dev --no-install-project --python 3.12
+
+# 4. 拷贝项目文件并安装项目本身
+RUN uv sync --frozen --no-dev --python 3.12
+
+ENV PATH="/app/.venv/bin:$PATH"
+
+EXPOSE 8000
+CMD ["uv", "run","-m","app.litserve_serve"]

app/litserve_serve.py

@@ -6,10 +6,9 @@ from minio import Minio
 import litserve as ls
 from PIL import Image
 import io
-import base64
 from diffusers import Flux2KleinPipeline
 
-from utils.new_oss_client import oss_get_image, oss_upload_image, MINIO_URL, MINIO_ACCESS, MINIO_SECRET, MINIO_SECURE
+from app.utils.new_oss_client import oss_get_image, oss_upload_image, MINIO_URL, MINIO_ACCESS, MINIO_SECRET, MINIO_SECURE
 
 minio_client = Minio(MINIO_URL, access_key=MINIO_ACCESS, secret_key=MINIO_SECRET, secure=MINIO_SECURE)
 
@@ -117,9 +116,10 @@ class FluxKleinAPI(ls.LitAPI):
         images = payload.get("images", [])
         prompt = payload.get("prompt", "")
         gen = torch.Generator(device=self.device)
-        output = {}
+        seed = gen.seed()
+        print(f"本次使用的随机种子是: {seed}")
         if images:
-            output['im'] = self.pipe(
+            output = self.pipe(
                 image=images,
                 prompt=prompt,
                 height=payload.get("height", 512),

app/utils/new_oss_client.py

@@ -61,7 +61,7 @@ def oss_upload_image(oss_client, bucket, object_name, image_bytes):
 
 
 if __name__ == '__main__':
-    url = "test/fida_generate_image/692932cfecd9499688899812737a369c.png"
+    url = "test/fida_generate_image/9fea0a597d884c81a5a4c8381b7bae2c.png"
     img = oss_get_image(oss_client=minio_client, bucket=url.split('/')[0], object_name=url[url.find('/') + 1:])
     img.show()
     img.save("result.png")

docker-compose.yml

@@ -0,0 +1,19 @@
+services:
+  flux2_klein:
+    restart: always
+    build:
+      context: .
+      dockerfile: Dockerfile
+    working_dir: /app
+    volumes:
+      - ./app:/app/app
+      - /etc/localtime:/etc/localtime:ro
+    ports:
+      - "14202:8451"
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              device_ids: [ '0' ]
+              capabilities: [ gpu ]

pyproject.toml

@@ -21,6 +21,7 @@ dependencies = [
   "datetime>=6.0",
   "litserve>=0.2.17",
   "minio>=7.2.20",
+  "diffusers",
 ]
 
 [project.optional-dependencies]
@@ -46,3 +47,6 @@ quote-style = "double"
 indent-style = "space"
 line-ending = "auto"
 docstring-code-format = true
+
+[tool.uv.sources]
+diffusers = { git = "https://github.com/huggingface/diffusers.git" }

test.py

@@ -1,31 +0,0 @@
-import datetime
-import time
-
-import torch
-from PIL import Image
-from diffusers import Flux2KleinPipeline
-
-device = "cuda"
-dtype = torch.bfloat16
-
-pipe = Flux2KleinPipeline.from_pretrained("black-forest-labs/FLUX.2-klein-4B", torch_dtype=dtype, is_distilled=False)
-pipe.to(device)  # save some VRAM by offloading the model to CPUsave some VRAM by offloading the model to CPU
-
-timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
-prompt = ""
-num_inference_steps = 4
-input_image = Image.open("result1.png")
-
-start_time = time.time()
-image = pipe(
-    image=input_image,
-    prompt=prompt,
-    height=768,
-    width=512,
-    guidance_scale=1.0,
-    num_inference_steps=num_inference_steps,
-    # generator=torch.Generator(device=device).manual_seed(3)
-).images[0]
-
-image.save(f"{timestamp}_{num_inference_steps}steps.png")
-print(f"infer time : {time.time() - start_time}")

uv.lock

@@ -244,6 +244,23 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/cf/7a/ea0f3e3ea74be36fc7cf54f966cde732a3de72697983cdb5646b0a4dacde/datetime-6.0-py3-none-any.whl", hash = "sha256:d19988f0657a4e72c9438344157254a8dcad6aea8cd5ae70a5d1b5a75e5dc930", size = 52637, upload-time = "2025-11-25T08:00:33.077Z" },
 ]
 
+[[package]]
+name = "diffusers"
+version = "0.38.0.dev0"
+source = { git = "https://github.com/huggingface/diffusers.git#11a3284cee024bfff017408b303d18e88981d2ea" }
+dependencies = [
+    { name = "filelock" },
+    { name = "httpx" },
+    { name = "huggingface-hub" },
+    { name = "importlib-metadata" },
+    { name = "numpy", version = "2.2.6", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.11'" },
+    { name = "numpy", version = "2.4.1", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.11'" },
+    { name = "pillow" },
+    { name = "regex" },
+    { name = "requests" },
+    { name = "safetensors" },
+]
+
 [[package]]
 name = "distro"
 version = "1.9.0"
@@ -317,6 +334,7 @@ source = { editable = "." }
 dependencies = [
     { name = "accelerate" },
     { name = "datetime" },
+    { name = "diffusers" },
     { name = "einops" },
     { name = "fire" },
     { name = "litserve" },
@@ -337,6 +355,7 @@ dev = [
 requires-dist = [
     { name = "accelerate", specifier = "==1.12.0" },
     { name = "datetime", specifier = ">=6.0" },
+    { name = "diffusers", git = "https://github.com/huggingface/diffusers.git" },
     { name = "einops", specifier = "==0.8.1" },
     { name = "fire", specifier = "==0.7.1" },
     { name = "litserve", specifier = ">=0.2.17" },
@@ -468,6 +487,18 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/0e/61/66938bbb5fc52dbdf84594873d5b51fb1f7c7794e9c0f5bd885f30bc507b/idna-3.11-py3-none-any.whl", hash = "sha256:771a87f49d9defaf64091e6e6fe9c18d4833f140bd19464795bc32d966ca37ea", size = 71008, upload-time = "2025-10-12T14:55:18.883Z" },
 ]
 
+[[package]]
+name = "importlib-metadata"
+version = "8.7.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "zipp" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/f3/49/3b30cad09e7771a4982d9975a8cbf64f00d4a1ececb53297f1d9a7be1b10/importlib_metadata-8.7.1.tar.gz", hash = "sha256:49fef1ae6440c182052f407c8d34a68f72efc36db9ca90dc0113398f2fdde8bb", size = 57107, upload-time = "2025-12-21T10:00:19.278Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/fa/5e/f8e9a1d23b9c20a551a8a02ea3637b4642e22c2626e3a13a9a29cdea99eb/importlib_metadata-8.7.1-py3-none-any.whl", hash = "sha256:5a1f80bf1daa489495071efbb095d75a634cf28a8bc299581244063b53176151", size = 27865, upload-time = "2025-12-21T10:00:18.329Z" },
+]
+
 [[package]]
 name = "jinja2"
 version = "3.1.6"
@@ -1697,6 +1728,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/6f/28/258ebab549c2bf3e64d2b0217b973467394a9cea8c42f70418ca2c5d0d2e/websockets-16.0-py3-none-any.whl", hash = "sha256:1637db62fad1dc833276dded54215f2c7fa46912301a24bd94d45d46a011ceec", size = 171598, upload-time = "2026-01-10T09:23:45.395Z" },
 ]
 
+[[package]]
+name = "zipp"
+version = "3.23.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/e3/02/0f2892c661036d50ede074e376733dca2ae7c6eb617489437771209d4180/zipp-3.23.0.tar.gz", hash = "sha256:a07157588a12518c9d4034df3fbbee09c814741a33ff63c05fa29d26a2404166", size = 25547, upload-time = "2025-06-08T17:06:39.4Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/2e/54/647ade08bf0db230bfea292f893923872fd20be6ac6f53b2b936ba839d75/zipp-3.23.0-py3-none-any.whl", hash = "sha256:071652d6115ed432f5ce1d34c336c0adfd6a884660d1e9712a256d3d3bd4b14e", size = 10276, upload-time = "2025-06-08T17:06:38.034Z" },
+]
+
 [[package]]
 name = "zope-interface"
 version = "8.2"