format

tests/models/test_compressed.py

@@ -9,6 +9,7 @@
 import gc
 
 import pytest
+
 from tests.models.utils import check_logprobs_close
 
 MAX_MODEL_LEN = 1024

tests/models/test_gptq_marlin.py

@@ -13,11 +13,10 @@
 import pytest
 import torch
 
+from tests.models.utils import check_logprobs_close
 from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
 from vllm.model_executor.layers.rotary_embedding import _ROPE_DICT
 
-from tests.models.utils import check_logprobs_close
-
 os.environ["TOKENIZERS_PARALLELISM"] = "true"
 
 MAX_MODEL_LEN = 1024

tests/models/test_marlin.py

@@ -20,9 +20,8 @@
 import pytest
 import torch
 
-from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
-
 from tests.models.utils import check_logprobs_close
+from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
 
 capability = torch.cuda.get_device_capability()
 capability = capability[0] * 10 + capability[1]

tests/models/test_models_logprobs.py

@@ -3,6 +3,7 @@
 Run `pytest tests/models/test_models_logprobs.py --forked`.
 """
 import pytest
+
 from tests.models.utils import check_logprobs_close
 
 MODEL_MAX_LEN = 1024

tests/tensorizer_loader/test_tensorizer.py

@@ -2,14 +2,14 @@
 import json
 import os
 import subprocess
-from tests.utils import ServerRunner
 from unittest.mock import MagicMock, patch
 
 import openai
 import pytest
 import ray
 import torch
 
+from tests.utils import ServerRunner
 from vllm import SamplingParams
 # yapf: disable
 from vllm.model_executor.model_loader.tensorizer import (TensorizerConfig,