From 8982ff3b1c1926916465270fd62139f33d812347 Mon Sep 17 00:00:00 2001 From: junq <22017000+QiJune@users.noreply.github.com> Date: Thu, 28 Aug 2025 16:04:59 +0800 Subject: [PATCH 1/2] fix mocker.patch.object error Signed-off-by: junq <22017000+QiJune@users.noreply.github.com> --- tests/integration/defs/accuracy/test_llm_api_pytorch.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py index cce443bf7ba5..d6a3884ee14d 100644 --- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py +++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py @@ -2736,7 +2736,7 @@ def test_w4_1gpu(self, moe_backend, cuda_graph, overlap_scheduler, mocker): with llm: model_name = "GPT-OSS/MXFP4" - mocker.patch.object(GSM8K, {"MAX_OUTPUT_LEN": 8192}) + mocker.patch.object(GSM8K, "MAX_OUTPUT_LEN", 8192) task = GSM8K(model_name) task.evaluate(llm, extra_evaluator_kwargs=self.extra_evaluator_kwargs) @@ -2776,7 +2776,7 @@ def test_w4_4gpus(self, moe_backend, tp_size, pp_size, ep_size, with llm: model_name = "GPT-OSS/MXFP4" task = GSM8K(model_name) - mocker.patch.object(GSM8K, {"MAX_OUTPUT_LEN": 8192}) + mocker.patch.object(GSM8K, "MAX_OUTPUT_LEN", 8192) task.evaluate(llm, extra_evaluator_kwargs=self.extra_evaluator_kwargs) From 72bf1607c7a21a93320d046985d5c8afe0fb3490 Mon Sep 17 00:00:00 2001 From: junq <22017000+QiJune@users.noreply.github.com> Date: Thu, 28 Aug 2025 16:40:38 +0800 Subject: [PATCH 2/2] skip TestGPTOSS Signed-off-by: junq <22017000+QiJune@users.noreply.github.com> --- tests/integration/defs/accuracy/test_llm_api_pytorch.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py index d6a3884ee14d..d1f3999fbab7 100644 --- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py +++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py @@ -2719,6 +2719,7 @@ class TestGPTOSS(LlmapiAccuracyTestHarness): (True, True), ]) def test_w4_1gpu(self, moe_backend, cuda_graph, overlap_scheduler, mocker): + pytest.skip("https://nvbugs/5481087") if moe_backend == "TRITON" and not IS_TRITON_KERNELS_AVAILABLE: pytest.skip("Triton kernels are not available") @@ -2756,6 +2757,7 @@ def test_w4_1gpu(self, moe_backend, cuda_graph, overlap_scheduler, mocker): ids=["tp4", "ep4", "dp4"]) def test_w4_4gpus(self, moe_backend, tp_size, pp_size, ep_size, attention_dp, cuda_graph, overlap_scheduler, mocker): + pytest.skip("https://nvbugs/5481087") if moe_backend == "TRITON": if not IS_TRITON_KERNELS_AVAILABLE: pytest.skip("Triton kernels are not available")