[Misc] Add vLLM version getter to utils (#5098)

vllm/engine/llm_engine.py

@@ -6,7 +6,6 @@ from typing import Type, TypeVar, Union
 
 from transformers import GenerationConfig, PreTrainedTokenizer
 
-import vllm
 from vllm.config import (CacheConfig, DecodingConfig, DeviceConfig, LoadConfig,
                          LoRAConfig, ModelConfig, ParallelConfig,
                          SchedulerConfig, SpeculativeConfig,
@@ -38,6 +37,7 @@ from vllm.transformers_utils.tokenizer_group import (BaseTokenizerGroup,
 from vllm.usage.usage_lib import (UsageContext, is_usage_stats_enabled,
                                   usage_message)
 from vllm.utils import Counter
+from vllm.version import __version__ as VLLM_VERSION
 
 logger = init_logger(__name__)
 _LOCAL_LOGGING_INTERVAL_SEC = 5
@@ -169,7 +169,7 @@ class LLMEngine:
             "enforce_eager=%s, kv_cache_dtype=%s, "
             "quantization_param_path=%s, device_config=%s, "
             "decoding_config=%r, seed=%d, served_model_name=%s)",
-            vllm.__version__,
+            VLLM_VERSION,
             model_config.model,
             speculative_config,
             model_config.tokenizer,