[NPU] Add MiMo-V2-Flash manual testcases (#28223)

test/manual/ascend/llm_models/test_npu_mimo_v2_flash.py

@@ -0,0 +1,39 @@
+import unittest
+
+from sglang.test.ascend.gsm8k_ascend_mixin import GSM8KAscendMixin
+from sglang.test.ascend.test_ascend_utils import MIMO_V2_FLASH_WEIGHTS_PATH
+from sglang.test.test_utils import CustomTestCase
+
+
+class TestMiMoV2FlashGraphWithMTP(GSM8KAscendMixin, CustomTestCase):
+    """Testcase: Verify the inference accuracy of MiMo-V2-Flash on GSM8K with cuda graph and MTP (speculative decoding).
+
+    [Test Category] Model
+    [Test Target] XiaomiMiMo/MiMo-V2-Flash
+    [Test Config] Prefill+Decode, cuda graph enabled, EAGLE speculative decoding
+    """
+
+    model = MIMO_V2_FLASH_WEIGHTS_PATH
+    accuracy = 0.9
+    other_args = [
+        "--trust-remote-code",
+        "--mem-fraction-static",
+        "0.8",
+        "--attention-backend",
+        "ascend",
+        "--tp-size",
+        "16",
+        "--speculative-algorithm",
+        "EAGLE",
+        "--speculative-num-steps",
+        "3",
+        "--speculative-eagle-topk",
+        "1",
+        "--speculative-num-draft-tokens",
+        "4",
+        "--enable-multi-layer-eagle",
+    ]
+
+
+if __name__ == "__main__":
+    unittest.main()