[feat] Add Vertex AI compatible prediction route for /generate (#3866)

test/srt/run_suite.py

@@ -50,6 +50,7 @@ suites = {
         "test_hidden_states.py",
         "test_update_weights_from_disk.py",
         "test_update_weights_from_tensor.py",
+        "test_vertex_endpoint.py",
         "test_vision_chunked_prefill.py",
         "test_vision_llm.py",
         "test_vision_openai_server.py",

test/srt/test_vertex_endpoint.py

@@ -0,0 +1,52 @@
+"""
+python3 -m unittest test_vertex_endpoint.TestVertexEndpoint.test_vertex_generate
+"""
+
+import unittest
+
+import requests
+
+from sglang.srt.utils import kill_process_tree
+from sglang.test.test_utils import (
+    DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
+    DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+    DEFAULT_URL_FOR_TEST,
+    popen_launch_server,
+)
+
+
+class TestVertexEndpoint(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.model = DEFAULT_SMALL_MODEL_NAME_FOR_TEST
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        kill_process_tree(cls.process.pid)
+
+    def run_generate(self, parameters):
+        data = {
+            "instances": [
+                {"text": "The capital of France is"},
+                {"text": "The capital of China is"},
+            ],
+            "parameters": parameters,
+        }
+        response = requests.post(self.base_url + "/vertex_generate", json=data)
+        response_json = response.json()
+        assert len(response_json["predictions"]) == len(data["instances"])
+        return response_json
+
+    def test_vertex_generate(self):
+        for parameters in [None, {"sampling_params": {"max_new_tokens": 4}}]:
+            self.run_generate(parameters)
+
+
+if __name__ == "__main__":
+    unittest.main()