[CI] Try introducing isort. (#3495)

vllm/attention/__init__.py

-from vllm.attention.backends.abstract import AttentionBackend, AttentionMetadata
+from vllm.attention.backends.abstract import (AttentionBackend,
+                                              AttentionMetadata)
 from vllm.attention.layer import Attention
 from vllm.attention.selector import get_attn_backend
 

vllm/attention/backends/flash_attn.py

@@ -7,12 +7,13 @@ flashinfer for all the attention operations.
 from dataclasses import dataclass
 from typing import Dict, List, Optional, Tuple, Type
 
-from flash_attn import flash_attn_varlen_func
 import torch
+from flash_attn import flash_attn_varlen_func
 
 from vllm.attention.backends.abstract import (AttentionBackend, AttentionImpl,
                                               AttentionMetadata)
-from vllm.attention.ops.paged_attn import PagedAttention, PagedAttentionMetadata
+from vllm.attention.ops.paged_attn import (PagedAttention,
+                                           PagedAttentionMetadata)
 
 
 class FlashAttentionBackend(AttentionBackend):

vllm/attention/backends/xformers.py

@@ -11,7 +11,8 @@ from xformers.ops.fmha.attn_bias import (AttentionBias,
 
 from vllm.attention.backends.abstract import (AttentionBackend, AttentionImpl,
                                               AttentionMetadata)
-from vllm.attention.ops.paged_attn import PagedAttention, PagedAttentionMetadata
+from vllm.attention.ops.paged_attn import (PagedAttention,
+                                           PagedAttentionMetadata)
 from vllm.logger import init_logger
 from vllm.utils import is_hip
 

vllm/attention/ops/paged_attn.py

@@ -3,8 +3,7 @@ from typing import Dict, List, Optional, Tuple
 
 import torch
 
-from vllm._C import cache_ops
-from vllm._C import ops
+from vllm._C import cache_ops, ops
 from vllm.attention.ops.prefix_prefill import context_attention_fwd
 
 # Should be the same as PARTITION_SIZE in `paged_attention_v2_launcher`.

vllm/attention/selector.py

@@ -13,11 +13,13 @@ logger = init_logger(__name__)
 def get_attn_backend(dtype: torch.dtype) -> AttentionBackend:
     if _can_use_flash_attn(dtype):
         logger.info("Using FlashAttention backend.")
-        from vllm.attention.backends.flash_attn import FlashAttentionBackend  # noqa: F401
+        from vllm.attention.backends.flash_attn import (  # noqa: F401
+            FlashAttentionBackend)
         return FlashAttentionBackend
     else:
         logger.info("Using XFormers backend.")
-        from vllm.attention.backends.xformers import XFormersBackend  # noqa: F401
+        from vllm.attention.backends.xformers import (  # noqa: F401
+            XFormersBackend)
         return XFormersBackend
 
 

vllm/config.py

-from typing import TYPE_CHECKING, Optional, Union, ClassVar
-from dataclasses import dataclass
+import json
 import os
-from packaging.version import Version
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, ClassVar, Optional, Union
 
-import json
 import torch
+from packaging.version import Version
 from transformers import PretrainedConfig
 
 from vllm.logger import init_logger
 from vllm.transformers_utils.config import get_config
-from vllm.utils import get_cpu_memory, is_hip, is_neuron, get_nvcc_cuda_version
+from vllm.utils import get_cpu_memory, get_nvcc_cuda_version, is_hip, is_neuron
 
 if TYPE_CHECKING:
     from ray.util.placement_group import PlacementGroup
@@ -103,7 +103,8 @@ class ModelConfig:
         if os.environ.get("VLLM_USE_MODELSCOPE", "False").lower() == "true":
             # download model from ModelScope hub,
             # lazy import so that modelscope is not required for normal use.
-            from modelscope.hub.snapshot_download import snapshot_download  # pylint: disable=C
+            # pylint: disable=C.
+            from modelscope.hub.snapshot_download import snapshot_download
 
             if not os.path.exists(model):
                 model_path = snapshot_download(model_id=model,

vllm/core/block_manager.py

 """A block manager that manages token blocks."""
 import enum
+from abc import ABC, abstractmethod
 from itertools import count, takewhile
 from os.path import commonprefix
 from typing import Dict, List, Optional, Set, Tuple
-from abc import ABC, abstractmethod
 
 from vllm.block import BlockTable, PhysicalTokenBlock
+from vllm.core.evictor import EvictionPolicy, Evictor, make_evictor
+from vllm.logger import init_logger
 from vllm.sequence import Sequence, SequenceGroup, SequenceStatus
 from vllm.utils import Device
-from vllm.core.evictor import Evictor, EvictionPolicy, make_evictor
-from vllm.logger import init_logger
 
 logger = init_logger(__name__)
 

vllm/core/evictor.py

 import enum
-from typing import OrderedDict
 from abc import ABC, abstractmethod, abstractproperty
+from typing import OrderedDict
 
 from vllm.block import PhysicalTokenBlock
 

vllm/core/scheduler.py

-from collections import deque
 import enum
 import time
-from typing import Deque, Dict, Iterable, List, Optional, Tuple, Union, Set
+from collections import deque
+from typing import Deque, Dict, Iterable, List, Optional, Set, Tuple, Union
 
 from vllm.config import CacheConfig, LoRAConfig, SchedulerConfig
 from vllm.core.block_manager import AllocStatus, BlockSpaceManager
 from vllm.core.policy import PolicyFactory
-from vllm.lora.request import LoRARequest
 from vllm.logger import init_logger
+from vllm.lora.request import LoRARequest
 from vllm.sequence import (Sequence, SequenceData, SequenceGroup,
                            SequenceGroupMetadata, SequenceStatus)
 

vllm/engine/arg_utils.py

@@ -3,9 +3,8 @@ import dataclasses
 from dataclasses import dataclass
 from typing import Optional, Tuple
 
-from vllm.config import (CacheConfig, DeviceConfig, ModelConfig,
-                         ParallelConfig, SchedulerConfig, LoRAConfig,
-                         TokenizerPoolConfig)
+from vllm.config import (CacheConfig, DeviceConfig, LoRAConfig, ModelConfig,
+                         ParallelConfig, SchedulerConfig, TokenizerPoolConfig)
 
 
 @dataclass

vllm/engine/async_llm_engine.py

@@ -2,17 +2,17 @@ import asyncio
 import os
 import time
 from functools import partial
-from typing import (Callable, Dict, Iterable, List, Optional, Set, Tuple, Type,
-                    Union, AsyncIterator)
+from typing import (AsyncIterator, Callable, Dict, Iterable, List, Optional,
+                    Set, Tuple, Type, Union)
 
 from transformers import PreTrainedTokenizer
 
-from vllm.lora.request import LoRARequest
 from vllm.config import ModelConfig
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.llm_engine import LLMEngine
 from vllm.engine.ray_utils import initialize_ray_cluster, ray
 from vllm.logger import init_logger
+from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
 from vllm.sampling_params import SamplingParams
 

vllm/engine/llm_engine.py

@@ -4,22 +4,22 @@ from typing import Iterable, List, Optional, Tuple, Type, Union
 from transformers import PreTrainedTokenizer
 
 import vllm
-from vllm.lora.request import LoRARequest
-from vllm.config import (CacheConfig, DeviceConfig, ModelConfig,
-                         ParallelConfig, SchedulerConfig, LoRAConfig)
+from vllm.config import (CacheConfig, DeviceConfig, LoRAConfig, ModelConfig,
+                         ParallelConfig, SchedulerConfig)
 from vllm.core.scheduler import Scheduler, SchedulerOutputs
 from vllm.engine.arg_utils import EngineArgs
-from vllm.executor.executor_base import ExecutorBase
 from vllm.engine.metrics import StatLogger, Stats
 from vllm.engine.ray_utils import initialize_ray_cluster
+from vllm.executor.executor_base import ExecutorBase
 from vllm.logger import init_logger
+from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
 from vllm.sampling_params import SamplingParams
 from vllm.sequence import (SamplerOutput, Sequence, SequenceGroup,
                            SequenceGroupOutput, SequenceOutput, SequenceStatus)
+from vllm.transformers_utils.detokenizer import Detokenizer
 from vllm.transformers_utils.tokenizer_group import (BaseTokenizerGroup,
                                                      get_tokenizer_group)
-from vllm.transformers_utils.detokenizer import Detokenizer
 from vllm.utils import Counter
 
 logger = init_logger(__name__)

vllm/engine/metrics.py

-from vllm.logger import init_logger
-from prometheus_client import (Counter, Gauge, Histogram, Info, REGISTRY,
-                               disable_created_metrics)
-
 import time
-import numpy as np
-from typing import Dict, List
 from dataclasses import dataclass
+from typing import Dict, List
+
+import numpy as np
+from prometheus_client import (REGISTRY, Counter, Gauge, Histogram, Info,
+                               disable_created_metrics)
+
+from vllm.logger import init_logger
 
 logger = init_logger(__name__)
 

vllm/engine/ray_utils.py

 import pickle
-
-from typing import Optional, List, Tuple
+from typing import List, Optional, Tuple
 
 from vllm.config import ParallelConfig
 from vllm.logger import init_logger
-from vllm.utils import is_hip, set_cuda_visible_devices, get_ip
+from vllm.utils import get_ip, is_hip, set_cuda_visible_devices
 
 logger = init_logger(__name__)
 

vllm/entrypoints/api_server.py

@@ -11,9 +11,9 @@ import json
 import ssl
 from typing import AsyncGenerator
 
+import uvicorn
 from fastapi import FastAPI, Request
 from fastapi.responses import JSONResponse, Response, StreamingResponse
-import uvicorn
 
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.async_llm_engine import AsyncLLMEngine

vllm/entrypoints/llm.py

@@ -3,9 +3,9 @@ from typing import List, Optional, Union
 from tqdm import tqdm
 from transformers import PreTrainedTokenizer, PreTrainedTokenizerFast
 
-from vllm.lora.request import LoRARequest
 from vllm.engine.arg_utils import EngineArgs
 from vllm.engine.llm_engine import LLMEngine
+from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
 from vllm.sampling_params import SamplingParams
 from vllm.utils import Counter

vllm/entrypoints/openai/api_server.py

 import asyncio
-from contextlib import asynccontextmanager
-import os
 import importlib
 import inspect
+import os
+from contextlib import asynccontextmanager
+from http import HTTPStatus
 
-from prometheus_client import make_asgi_app
 import fastapi
 import uvicorn
-from http import HTTPStatus
 from fastapi import Request
 from fastapi.exceptions import RequestValidationError
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import JSONResponse, StreamingResponse, Response
+from fastapi.responses import JSONResponse, Response, StreamingResponse
+from prometheus_client import make_asgi_app
 
 import vllm
 from vllm.engine.arg_utils import AsyncEngineArgs
 from vllm.engine.async_llm_engine import AsyncLLMEngine
-from vllm.entrypoints.openai.protocol import (CompletionRequest,
-                                              ChatCompletionRequest,
-                                              ErrorResponse)
-from vllm.logger import init_logger
 from vllm.entrypoints.openai.cli_args import make_arg_parser
+from vllm.entrypoints.openai.protocol import (ChatCompletionRequest,
+                                              CompletionRequest, ErrorResponse)
 from vllm.entrypoints.openai.serving_chat import OpenAIServingChat
 from vllm.entrypoints.openai.serving_completion import OpenAIServingCompletion
+from vllm.logger import init_logger
 
 TIMEOUT_KEEP_ALIVE = 5  # seconds
 

vllm/entrypoints/openai/protocol.py

@@ -3,12 +3,11 @@
 import time
 from typing import Dict, List, Literal, Optional, Union
 
+import torch
 from pydantic import BaseModel, Field, model_validator
 
-from vllm.utils import random_uuid
 from vllm.sampling_params import SamplingParams
-
-import torch
+from vllm.utils import random_uuid
 
 
 class ErrorResponse(BaseModel):

vllm/entrypoints/openai/serving_chat.py

-import time
 import codecs
+import time
+from typing import AsyncGenerator, AsyncIterator, List, Optional, Union
+
 from fastapi import Request
-from typing import AsyncGenerator, AsyncIterator, Optional, List, Union
-from vllm.logger import init_logger
-from vllm.utils import random_uuid
+
 from vllm.engine.async_llm_engine import AsyncLLMEngine
 from vllm.entrypoints.openai.protocol import (
     ChatCompletionRequest, ChatCompletionResponse,
     ChatCompletionResponseChoice, ChatCompletionResponseStreamChoice,
     ChatCompletionStreamResponse, ChatMessage, DeltaMessage, ErrorResponse,
     UsageInfo)
-from vllm.outputs import RequestOutput
-from vllm.entrypoints.openai.serving_engine import OpenAIServing, LoRA
+from vllm.entrypoints.openai.serving_engine import LoRA, OpenAIServing
+from vllm.logger import init_logger
 from vllm.model_executor.guided_decoding import (
     get_guided_decoding_logits_processor)
+from vllm.outputs import RequestOutput
+from vllm.utils import random_uuid
 
 logger = init_logger(__name__)
 

vllm/entrypoints/openai/serving_completion.py

 import asyncio
 import time
+from typing import (AsyncGenerator, AsyncIterator, Callable, Dict, List,
+                    Optional, Tuple)
+
 from fastapi import Request
-from typing import (AsyncGenerator, AsyncIterator, Callable, List, Optional,
-                    Dict, Tuple)
-from vllm.logger import init_logger
-from vllm.utils import random_uuid
+
 from vllm.engine.async_llm_engine import AsyncLLMEngine
-from vllm.entrypoints.openai.protocol import (
-    CompletionRequest,
-    CompletionResponse,
-    CompletionResponseChoice,
-    CompletionResponseStreamChoice,
-    CompletionStreamResponse,
-    LogProbs,
-    UsageInfo,
-)
-from vllm.outputs import RequestOutput
-from vllm.entrypoints.openai.serving_engine import OpenAIServing, LoRA
+from vllm.entrypoints.openai.protocol import (CompletionRequest,
+                                              CompletionResponse,
+                                              CompletionResponseChoice,
+                                              CompletionResponseStreamChoice,
+                                              CompletionStreamResponse,
+                                              LogProbs, UsageInfo)
+from vllm.entrypoints.openai.serving_engine import LoRA, OpenAIServing
+from vllm.logger import init_logger
 from vllm.model_executor.guided_decoding import (
     get_guided_decoding_logits_processor)
+from vllm.outputs import RequestOutput
+from vllm.utils import random_uuid
 
 logger = init_logger(__name__)