diff --git a/test/srt/models/test_grok_models.py b/test/srt/models/test_grok_models.py new file mode 100644 index 000000000..814f09a00 --- /dev/null +++ b/test/srt/models/test_grok_models.py @@ -0,0 +1,52 @@ +import unittest +from types import SimpleNamespace + +from sglang.srt.utils import kill_process_tree +from sglang.test.few_shot_gsm8k import run_eval +from sglang.test.test_utils import ( + DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH, + DEFAULT_URL_FOR_TEST, + popen_launch_server, +) + + +class TestGrok(unittest.TestCase): + @classmethod + def setUpClass(cls): + cls.model = "lmzheng/grok-1" + cls.base_url = DEFAULT_URL_FOR_TEST + cls.process = popen_launch_server( + cls.model, + cls.base_url, + timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH, + other_args=[ + "--load-format", + "dummy", + "--json-model-override-args", + '{"num_hidden_layers": 2}', + ], + ) + + @classmethod + def tearDownClass(cls): + kill_process_tree(cls.process.pid) + + def test_gsm8k(self): + args = SimpleNamespace( + num_shots=5, + data_path=None, + num_questions=64, + max_new_tokens=256, + parallel=128, + host="http://127.0.0.1", + port=int(self.base_url.split(":")[-1]), + ) + metrics = run_eval(args) + print(f"{metrics=}") + + # It is dummy weights so we only assert the output throughput instead of accuracy. + self.assertGreater(metrics["output_throughput"], 1000) + + +if __name__ == "__main__": + unittest.main() diff --git a/test/srt/run_suite.py b/test/srt/run_suite.py index 60ea570cb..ecd1a39dd 100644 --- a/test/srt/run_suite.py +++ b/test/srt/run_suite.py @@ -18,6 +18,7 @@ suites = { TestFile("models/lora/test_multi_lora_backend.py", 1), TestFile("models/test_embedding_models.py", 119), TestFile("models/test_generation_models.py", 103), + TestFile("models/test_grok_models.py", 60), TestFile("models/test_qwen_models.py", 82), TestFile("models/test_reward_models.py", 83), TestFile("models/test_gme_qwen_models.py", 45),