ci: migrate 1-GPU model tests to test/registered/models/ (#16414)

test/registered/models/test_vlm_models.py

@@ -0,0 +1,61 @@
+from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
+
+# VLM (Vision Language Model) tests
+register_cuda_ci(est_time=270, suite="stage-b-test-small-1-gpu")
+register_amd_ci(est_time=387, suite="stage-b-test-small-1-gpu")
+
+import argparse
+import random
+import sys
+import unittest
+from types import SimpleNamespace
+
+from sglang.srt.utils import is_hip
+from sglang.test.kits.mmmu_vlm_kit import (
+    DEFAULT_MEM_FRACTION_STATIC,
+    MMMUMultiModelTestBase,
+)
+from sglang.test.test_utils import is_in_ci
+
+_is_hip = is_hip()
+# VLM models for testing
+if _is_hip:
+    MODELS = [SimpleNamespace(model="openbmb/MiniCPM-V-2_6", mmmu_accuracy=0.4)]
+else:
+    MODELS = [
+        SimpleNamespace(model="google/gemma-3-4b-it", mmmu_accuracy=0.38),
+        SimpleNamespace(model="Qwen/Qwen2.5-VL-3B-Instruct", mmmu_accuracy=0.4),
+        SimpleNamespace(model="openbmb/MiniCPM-V-2_6", mmmu_accuracy=0.4),
+    ]
+
+
+class TestVLMModels(MMMUMultiModelTestBase):
+    def test_vlm_mmmu_benchmark(self):
+        """Test VLM models against MMMU benchmark."""
+        models_to_test = MODELS
+
+        if is_in_ci():
+            models_to_test = [random.choice(MODELS)]
+
+        for model in models_to_test:
+            self._run_vlm_mmmu_test(model, "./logs")
+
+
+if __name__ == "__main__":
+    # Define and parse arguments here, before unittest.main
+    parser = argparse.ArgumentParser(description="Test VLM models")
+    parser.add_argument(
+        "--mem-fraction-static",
+        type=float,
+        help="Static memory fraction for the model",
+        default=DEFAULT_MEM_FRACTION_STATIC,
+    )
+
+    # Parse args intended for unittest
+    args = parser.parse_args()
+
+    # Store the parsed args object on the class
+    TestVLMModels.parsed_args = args
+
+    # Pass args to unittest
+    unittest.main(argv=[sys.argv[0]])