[Tiny]Move deepseek fp4 cutlass moe test to per-commit test (#15565)

2025-12-21 18:08:07 -08:00
parent 796969caf0
commit 468931b572
4 changed files with 60 additions and 77 deletions
@@ -1,3 +1,4 @@
+import os
 import unittest
 from types import SimpleNamespace

@@ -149,5 +150,62 @@ class TestDeepseekV3FP4PiecewiseCudaGraph(CustomTestCase):
        self.assertGreater(speed, 120)


+class TestDeepseekV3FP4CutlassMoE(CustomTestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.model = FULL_DEEPSEEK_V3_FP4_MODEL_PATH
+        cls.base_url = DEFAULT_URL_FOR_TEST
+        other_args = [
+            "--tp",
+            "4",
+            "--ep",
+            "4",
+            "--attention-backend",
+            "trtllm_mla",
+            "--moe-runner-backend",
+            "flashinfer_cutlass",
+            "--quantization",
+            "modelopt_fp4",
+            "--model-loader-extra-config",
+            '{"enable_multithread_load": true}',
+        ]
+        cls.process = popen_launch_server(
+            cls.model,
+            cls.base_url,
+            timeout=SERVER_LAUNCH_TIMEOUT,
+            other_args=other_args,
+            env={
+                **os.environ,
+                "SGLANG_MOE_NVFP4_DISPATCH": "1",  # Enable nvfp4 all gather
+            },
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        kill_process_tree(cls.process.pid)
+
+    def test_a_gsm8k(
+        self,
+    ):  # Append an "a" to make this test run first (alphabetically) to warm up the server
+        args = SimpleNamespace(
+            num_shots=8,
+            data_path=None,
+            num_questions=1319,
+            parallel=1319,
+            max_new_tokens=512,
+            host="http://127.0.0.1",
+            port=int(self.base_url.split(":")[-1]),
+        )
+        metrics = run_eval_few_shot_gsm8k(args)
+        print(f"{metrics=}")
+
+        if is_in_ci():
+            write_github_step_summary(
+                f"### test_gsm8k (deepseek-v3-fp4-cutlass-moe)\n"
+                f'{metrics["accuracy"]=:.3f}\n'
+            )
+            self.assertGreater(metrics["accuracy"], 0.935)
+
+
 if __name__ == "__main__":
    unittest.main()