[3/N] CI refactor: move some manually triggered tests. (#13448)

docs/advanced_features/vlm_query.ipynb

@@ -49,7 +49,7 @@
     "image = Image.open(\n",
     "    BytesIO(\n",
     "        requests.get(\n",
-    "            \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "            \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "        ).content\n",
     "    )\n",
     ")\n",
@@ -187,7 +187,7 @@
     "image = Image.open(\n",
     "    BytesIO(\n",
     "        requests.get(\n",
-    "            \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "            \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "        ).content\n",
     "    )\n",
     ")\n",

docs/basic_usage/openai_api_vision.ipynb

@@ -75,7 +75,7 @@
     "          {{\n",
     "            \"type\": \"image_url\",\n",
     "            \"image_url\": {{\n",
-    "              \"url\": \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "              \"url\": \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "            }}\n",
     "          }}\n",
     "        ]\n",
@@ -120,7 +120,7 @@
     "                {\n",
     "                    \"type\": \"image_url\",\n",
     "                    \"image_url\": {\n",
-    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "                    },\n",
     "                },\n",
     "            ],\n",
@@ -163,7 +163,7 @@
     "                {\n",
     "                    \"type\": \"image_url\",\n",
     "                    \"image_url\": {\n",
-    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "                    },\n",
     "                },\n",
     "            ],\n",
@@ -203,7 +203,7 @@
     "                {\n",
     "                    \"type\": \"image_url\",\n",
     "                    \"image_url\": {\n",
-    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\",\n",
+    "                        \"url\": \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\",\n",
     "                    },\n",
     "                },\n",
     "                {\n",

docs/basic_usage/qwen3_vl.md

@@ -56,7 +56,7 @@ data = {
                 {
                     "type": "image_url",
                     "image_url": {
-                        "url": "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
+                        "url": "https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true"
                     },
                 },
             ],

docs/basic_usage/sampling_params.md

@@ -162,7 +162,7 @@ python3 -m sglang.launch_server --model-path lmms-lab/llava-onevision-qwen2-7b-o
 Download an image:
 
 ```bash
-curl -o example_image.png -L https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true
+curl -o example_image.png -L https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true
 ```
 
 Send a request:

docs/references/frontend/frontend_tutorial.ipynb

@@ -430,7 +430,7 @@
     "    s += assistant(gen(\"answer\", max_tokens=256))\n",
     "\n",
     "\n",
-    "image_url = \"https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true\"\n",
+    "image_url = \"https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true\"\n",
     "image_bytes, _ = load_image(image_url)\n",
     "state = image_qa(image_bytes, \"What is in the image?\")\n",
     "print_highlight(state[\"answer\"])"

examples/assets/.gitignore

@@ -0,0 +1 @@
+!example_image.png

examples/runtime/engine/offline_batch_inference_vlm.py

@@ -19,7 +19,7 @@ def main(
     conv = chat_templates[server_args.chat_template].copy()
     image_token = conv.image_token
 
-    image_url = "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
+    image_url = "https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true"
 
     prompt = f"What's in this image?\n{image_token}"
 

examples/runtime/multimodal/llava_onevision_server.py

@@ -98,7 +98,7 @@ def multi_image_stream_request_test(client):
                     {
                         "type": "image_url",
                         "image_url": {
-                            "url": "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png"
+                            "url": "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png"
                         },
                         "modalities": "multi-images",
                     },

python/sglang/test/send_one.py

@@ -78,17 +78,17 @@ def send_one_prompt(args):
         args.prompt = (
             "Human: Describe this image in a very short sentence.\n\nAssistant:"
         )
-        image_data = "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png"
+        image_data = "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png"
     elif args.many_images:
         args.prompt = (
             "Human: I have one reference image and many images."
             "Describe their relationship in a very short sentence.\n\nAssistant:"
         )
         image_data = [
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
         ]
     else:
         image_data = None

python/sglang/test/test_utils.py

@@ -123,7 +123,7 @@ DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_QUANT_TP1 = "hugging-quants/Meta-Llama-3.1-8
 DEFAULT_SMALL_MODEL_NAME_FOR_TEST_QWEN = "Qwen/Qwen2.5-1.5B-Instruct"
 DEFAULT_SMALL_VLM_MODEL_NAME_FOR_TEST = "Qwen/Qwen2.5-VL-3B-Instruct"
 
-DEFAULT_IMAGE_URL = "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
+DEFAULT_IMAGE_URL = "https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true"
 DEFAULT_VIDEO_URL = "https://raw.githubusercontent.com/EvolvingLMMs-Lab/sglang/dev/onevision_local/assets/jobs.mp4"
 
 DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH = 600

test/lang/test_litellm_backend.py

@@ -1,24 +0,0 @@
-import unittest
-
-from sglang import LiteLLM, set_default_backend
-from sglang.test.test_programs import test_mt_bench, test_stream
-from sglang.test.test_utils import CustomTestCase
-
-
-class TestAnthropicBackend(CustomTestCase):
-    chat_backend = None
-
-    @classmethod
-    def setUpClass(cls):
-        cls.chat_backend = LiteLLM("gpt-3.5-turbo")
-        set_default_backend(cls.chat_backend)
-
-    def test_mt_bench(self):
-        test_mt_bench()
-
-    def test_stream(self):
-        test_stream()
-
-
-if __name__ == "__main__":
-    unittest.main()

test/srt/run_suite.py

@@ -235,13 +235,6 @@ suites = {
         TestFile("debug_utils/test_log_parser.py", 5),
         TestFile("test_deepseek_v3_cutedsl_4gpu.py"),
         TestFile("entrypoints/http_server/test_abort_request.py"),
-        TestFile("ep/test_deepep_internode.py"),
-        TestFile("ep/test_deepep_intranode.py"),
-        TestFile("ep/test_deepep_low_latency.py"),
-        TestFile("ep/test_eplb.py"),
-        TestFile("ep/test_hybrid_dp_ep_tp_mtp.py"),
-        TestFile("ep/test_moe_deepep.py"),
-        TestFile("ep/test_moe_deepep_eval_accuracy_large.py"),
         TestFile("hicache/test_disaggregation_hicache.py"),
         TestFile("layers/attention/nsa/test_act_quant_triton.py"),
         TestFile("layers/moe/test_moe_runners.py"),

test/srt/test_session_control.py

@@ -594,8 +594,8 @@ class TestSessionControlVision(CustomTestCase):
             "<|im_start|>user\nDescribe this image in a very short sentence.<|im_end|>\nassistant:",
         ]
         image_chunks = [
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
-            "https://raw.githubusercontent.com/sgl-project/sglang/main/test/lang/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
+            "https://raw.githubusercontent.com/sgl-project/sglang/main/examples/assets/example_image.png",
             "https://raw.githubusercontent.com/sgl-project/sglang/main/assets/logo.png",
         ]
 

test/srt/test_vlm_accuracy.py

@@ -30,7 +30,7 @@ from sglang.srt.server_args import ServerArgs
 class VisionLLMLogitsBase(unittest.IsolatedAsyncioTestCase):
     @classmethod
     def setUpClass(cls):
-        cls.image_url = "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
+        cls.image_url = "https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true"
         cls.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
         cls.model_path = ""
         cls.chat_template = ""

test/srt/test_vlm_input_format.py

@@ -16,7 +16,7 @@ from sglang import Engine
 from sglang.srt.entrypoints.openai.protocol import ChatCompletionRequest
 from sglang.srt.parser.conversation import generate_chat_conv
 
-TEST_IMAGE_URL = "https://github.com/sgl-project/sglang/blob/main/test/lang/example_image.png?raw=true"
+TEST_IMAGE_URL = "https://github.com/sgl-project/sglang/blob/main/examples/assets/example_image.png?raw=true"
 
 
 class VLMInputTestBase: