add qwen3-next CI test of accuracy on BF16 and NVFP4

Signed-off-by: jiant <[email protected]>

Author: JadoTu

tests/integration/defs/accuracy/references/gsm8k.yaml

@@ -138,6 +138,11 @@ Qwen3/Qwen3-235B-A22B:
     accuracy: 85.78
 Qwen3/Qwen3-Next-80B-A3B-Thinking:
   - accuracy: 81.577
+Qwen3/Qwen3-Next-80B-A3B-Instruct:
+  - accuracy: 84.42
+  - quant_algo: NVFP4
+    kv_cache_quant_algo: FP8
+    accuracy: 84.32
 moonshotai/Kimi-K2-Instruct:
   - quant_algo: FP8_BLOCK_SCALES
     accuracy: 94.84

tests/integration/defs/accuracy/references/mmlu.yaml

@@ -242,6 +242,11 @@ Qwen3/Qwen3-235B-A22B:
     accuracy: 86
 Qwen3/Qwen3-Next-80B-A3B-Thinking:
   - accuracy: 86
+Qwen3/Qwen3-Next-80B-A3B-Instruct:
+  - accuracy: 85.58
+  - quant_algo: NVFP4
+    kv_cache_quant_algo: FP8
+    accuracy: 85
 moonshotai/Kimi-K2-Instruct:
   - quant_algo: FP8_BLOCK_SCALES
     accuracy: 87.65

tests/integration/defs/accuracy/test_llm_api_pytorch.py

@@ -4118,6 +4118,80 @@ def test_auto_dtype(self, tp_size, pp_size, ep_size):
             task.evaluate(llm)
 
 
+@skip_pre_hopper
+@pytest.mark.skip_less_device_memory(80000)
+class TestQwen3NextInstruct(LlmapiAccuracyTestHarness):
+    MODEL_PATH = f"{llm_models_root()}/Qwen3-Next"
+    MODEL_NAME = "Qwen3/Qwen3-Next-80B-A3B-Instruct"
+
+    @pytest.mark.skip_less_device(4)
+    @pytest.mark.parametrize(
+        "tp_size,pp_size,ep_size,cuda_graph,overlap_scheduler",
+        [
+            (4, 1, 4, True, True),
+        ],
+        ids=[
+            "tp4ep4_cudagraph_overlap",
+        ],
+    )
+    def test_bf16_4gpu(self, tp_size, pp_size, ep_size, cuda_graph,
+                       overlap_scheduler):
+        model_path = f"{self.MODEL_PATH}/Qwen3-Next-80B-A3B-Instruct"
+        model_path = "Qwen/Qwen3-Next-80B-A3B-Instruct"
+        kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6,
+                                        enable_block_reuse=False)
+        pytorch_config = dict(disable_overlap_scheduler=not overlap_scheduler,
+                              cuda_graph_config=CudaGraphConfig(
+                                  max_batch_size=512) if cuda_graph else None)
+
+        with LLM(
+                model_path,
+                tensor_parallel_size=tp_size,
+                max_num_tokens=16384,
+                pipeline_parallel_size=pp_size,
+                moe_expert_parallel_size=ep_size,
+                kv_cache_config=kv_cache_config,
+                **pytorch_config,
+        ) as llm:
+            task = MMLU(self.MODEL_NAME)
+            task.evaluate(llm)
+            task = GSM8K(self.MODEL_NAME)
+            task.evaluate(llm)
+
+    @skip_pre_blackwell
+    @pytest.mark.parametrize("moe_backend", ["CUTLASS", "TRTLLM"],
+                             ids=["cutlass", "trtllm"])
+    @pytest.mark.parametrize(
+        "tp_size,pp_size,ep_size,cuda_graph,overlap_scheduler",
+        [(1, 1, 1, True, True), (4, 1, 1, True, True), (4, 1, 4, True, True),
+         (4, 1, 4, False, False)],
+        ids=["tp1", "tp4ep1", "tp4ep4", "no_cuda_graph_overlap"])
+    def test_nvfp4(self, moe_backend, tp_size, pp_size, ep_size, cuda_graph,
+                   overlap_scheduler):
+        model_path = f"{self.MODEL_PATH}/qwen3-next-80b-instruct-nvfp4-ptq-fp8kv"
+        model_path = "/home/scratch.didow_sw_1/models/qwen3-next-80b-instruct-nvfp4-ptq-fp8kv"
+
+        kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.6,
+                                        enable_block_reuse=False)
+        pytorch_config = dict(disable_overlap_scheduler=not overlap_scheduler,
+                              cuda_graph_config=CudaGraphConfig(
+                                  max_batch_size=512) if cuda_graph else None)
+        moe_config = MoeConfig(backend=moe_backend)
+
+        with LLM(model_path,
+                 tensor_parallel_size=tp_size,
+                 max_num_tokens=16384,
+                 pipeline_parallel_size=pp_size,
+                 moe_expert_parallel_size=ep_size,
+                 kv_cache_config=kv_cache_config,
+                 **pytorch_config,
+                 moe_config=moe_config) as llm:
+            task = MMLU(self.MODEL_NAME)
+            task.evaluate(llm)
+            task = GSM8K(self.MODEL_NAME)
+            task.evaluate(llm)
+
+
 class TestSeedOss_36B(LlmapiAccuracyTestHarness):
     MODEL_NAME = "ByteDance-Seed/Seed-OSS-36B-Instruct"
     MODEL_PATH = f"{llm_models_root()}/Seed-OSS/Seed-OSS-36B-Instruct"

tests/integration/test_lists/test-db/l0_dgx_b200.yml

@@ -59,6 +59,12 @@ l0_dgx_b200:
   - accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_fp8_tp2pp2
   - accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_nvfp4_tp4
   - accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_4gpus[dp4-cutlass-auto]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_bf16_4gpu[tp4ep4_cudagraph_overlap]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp1-cutlass]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp4ep1-cutlass]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp4ep4-cutlass]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[no_cuda_graph_overlap-cutlass]
+  - accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_nvfp4[tp4ep4-trtllm]
 - condition:
     ranges:
       system_gpu_count: