[CI] xpu inference test

tests/test_inference_result_xpu.py

@@ -0,0 +1,85 @@
+# Copyright 2024-2025 ModelCloud.ai
+# Copyright 2024-2025 [email protected]
+# Contact: [email protected], x.com/qubitium
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+
+os.environ["CUDA_VISIBLE_DEVICES"] = ""
+import tempfile
+
+from gptqmodel import BACKEND, GPTQModel, QuantizeConfig
+from gptqmodel.models._const import DEVICE
+from models.model_test import ModelTest
+from parameterized import parameterized
+
+
+class TestInferenceResultXPU(ModelTest):
+    NATIVE_MODEL_ID = "/monster/data/model/Llama-3.2-1B-Instruct"
+
+    @parameterized.expand(
+        [
+            (BACKEND.TRITON, DEVICE.XPU, True),
+            (BACKEND.TRITON, DEVICE.XPU, False),
+            (BACKEND.TORCH, DEVICE.XPU, True),
+            (BACKEND.TORCH, DEVICE.XPU, False),
+        ]
+    )
+    def testTritonXPU(self, backend, device, template):
+        origin_model = GPTQModel.load(
+            self.NATIVE_MODEL_ID,
+            quantize_config=QuantizeConfig(),
+            backend=backend,
+            device=device,
+        )
+        tokenizer = self.load_tokenizer(self.NATIVE_MODEL_ID)
+        calibration_dataset = self.load_dataset(tokenizer, rows=128)
+        origin_model.quantize(calibration_dataset, backend=BACKEND.TRITON)
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+          origin_model.save(tmpdir)
+
+          messages = [
+              [{"role": "user", "content": "The capital of France is"}],
+              [{"role": "user", "content": "The capital of the United Kingdom is"}],
+              [{"role": "user", "content": "The largest ocean on Earth is"}],
+              [{"role": "user", "content": "The world’s longest river is"}],
+              [{"role": "user", "content": "The tallest mountain in the world is"}],
+              [{"role": "user", "content": "How are you?"}],
+              [{"role": "user", "content": "I love reading and ??."}],
+              [{"role": "user", "content": "What is the official language of China?"}],
+              [{"role": "user", "content": "I am a good ??."}],
+              [{"role": "user", "content": "What is the official language of France?"}],
+          ]
+
+          model = GPTQModel.load(
+              tmpdir,
+              backend=backend,
+              device=device,
+          )
+
+          tokenizer = model.tokenizer
+
+          for message in messages:
+              if template:
+                  inputs_tensor = tokenizer.apply_chat_template(
+                      message,
+                      add_generation_prompt=True,
+                      return_tensors='pt').to(model.device)
+              else:
+                  inputs_tensor = tokenizer(message[0]["content"], return_tensors="pt")["input_ids"].to(model.device)
+
+              result = model.generate(inputs_tensor, max_length=128, eos_token_id=tokenizer.eos_token_id)
+              generate_str = tokenizer.batch_decode(result[:, inputs_tensor.size(-1):],  skip_special_tokens=True)
+              print(f"generate_str: {generate_str}")