import unittest from types import SimpleNamespace from sglang.srt.utils import kill_process_tree from sglang.test.run_eval import run_eval from sglang.test.test_utils import ( DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH, DEFAULT_URL_FOR_TEST, CustomTestCase, popen_launch_server, ) class TestUnslothPhi4(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/phi-4" 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=[], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.78) class TestUnslothPhi4Bnb4bit(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/phi-4-bnb-4bit" 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", "bitsandbytes", ], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.75) class TestUnslothPhi4UnslothBnb4bit(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/phi-4-unsloth-bnb-4bit" 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", "bitsandbytes", ], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.75) class TestUnslothPhi4MiniInstruct(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/Phi-4-mini-instruct" 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=[], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.65) class TestUnslothPhi4MiniBnb4bit(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/Phi-4-mini-instruct-bnb-4bit" 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", "bitsandbytes", ], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.6) class TestUnslothPhi4MiniUnslothBnb4bit(CustomTestCase): @classmethod def setUpClass(cls): cls.model = "unsloth/Phi-4-mini-instruct-unsloth-bnb-4bit" 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", "bitsandbytes", ], ) @classmethod def tearDownClass(cls): kill_process_tree(cls.process.pid) def test_gsm8k(self): args = SimpleNamespace( base_url=self.base_url, model=self.model, eval_name="gsm8k", api="completion", max_tokens=512, num_examples=200, num_threads=128, ) metrics = run_eval(args) print(f"{metrics=}") self.assertGreater(metrics["score"], 0.6) if __name__ == "__main__": unittest.main()