Merge branch 'develop' into mm_structred_output

Author: Jiang-Jia-Jun

.github/workflows/ci.yml

@@ -2,7 +2,9 @@ name: CI
 
 on:
   pull_request:
-    branches: [ develop ]
+    branches: 
+      - develop
+      - 'release/*'
   workflow_dispatch:
 
 concurrency:

.github/workflows/ci_xpu.yml

@@ -2,7 +2,9 @@ name: CI_XPU
 
 on:
   pull_request:
-    branches: [ develop ]
+    branches: 
+      - develop
+      - 'release/*'
   workflow_dispatch:
 
 concurrency:

fastdeploy/entrypoints/openai/test_openai.py

@@ -21,7 +21,7 @@
 
 client = openai.Client(base_url=f"http://{ip}:{service_http_port}/v1", api_key="EMPTY_API_KEY")
 
-# 非流式返回
+# 非流式返回, completion接口不会使用chat template对输入进行处理
 response = client.completions.create(
     model="default",
     prompt="There are 50 kinds of fruits, include apple, banana, pineapple",
@@ -33,7 +33,7 @@
 print(response)
 print("\n")
 
-# 流式返回
+# 流式返回, completion接口不会使用chat template对输入进行处理
 response = client.completions.create(
     model="default",
     prompt="Hello, how are you?",
@@ -46,13 +46,12 @@
 print("\n")
 
 # Chat completion
-# 非流式返回
+# 非流式返回, 会基于chat template对输入进行拼接处理
 response = client.chat.completions.create(
     model="default",
     messages=[
-        {"role": "user", "content": "Hello, who are you"},
         {"role": "system", "content": "I'm a helpful AI assistant."},
-        {"role": "user", "content": "List 3 countries and their capitals."},
+        {"role": "user", "content": "Hello, who are you"},
     ],
     temperature=1,
     max_tokens=64,
@@ -63,13 +62,12 @@
 print("\n")
 
 
-# # 流式返回
+# # 流式返回, 会基于chat template对输入进行拼接处理
 response = client.chat.completions.create(
     model="default",
     messages=[
-        {"role": "user", "content": "Hello, who are you"},
         {"role": "system", "content": "I'm a helpful AI assistant."},
-        {"role": "user", "content": "List 3 countries and their capitals."},
+        {"role": "user", "content": "Hello, who are you"},
     ],
     temperature=1,
     max_tokens=64,

fastdeploy/input/mm_processor/image_preprocessor/image_preprocessor_adaptive.py

@@ -57,14 +57,6 @@
 MAX_PIXELS = 16384 * 28 * 28
 MAX_RATIO = 200
 
-VIDEO_MIN_PIXELS = 128 * 28 * 28
-VIDEO_MAX_PIXELS = 768 * 28 * 28
-VIDEO_TOTAL_PIXELS = 24576 * 28 * 28
-FRAME_FACTOR = 2
-FPS = 2.0
-FPS_MIN_FRAMES = 4
-FPS_MAX_FRAMES = 768
-
 
 VideoInput = Union[
     List["PIL.Image.Image"],