biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V0 Deprecation] Remove LLMEngine (#25033)

Browse Source 
Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>
Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
This commit is contained in:
Woosuk Kwon
2025-09-20 17:56:30 -07:00
committed by GitHub
parent 367a480bd3
commit 52c2a8d4ad
29 changed files with 65 additions and 2763 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/entrypoints/llm.py
View File

@@ -11,7 +11,6 @@ from pydantic import ValidationError
from tqdm.auto import tqdm
from typing_extensions import TypeVar
import vllm.envs as envs
from vllm.beam_search import (BeamSearchInstance, BeamSearchOutput,
BeamSearchSequence,
create_sort_beams_key_function)
@@ -19,7 +18,6 @@ from vllm.config import (CompilationConfig, ModelDType,
StructuredOutputsConfig, TokenizerMode, is_init_field)
from vllm.engine.arg_utils import (ConvertOption, EngineArgs, HfOverrides,
PoolerConfig, RunnerOption)
from vllm.engine.llm_engine import LLMEngine
from vllm.entrypoints.chat_utils import (ChatCompletionMessageParam,
ChatTemplateContentFormatOption,
apply_hf_chat_template,
@@ -54,6 +52,7 @@ from vllm.transformers_utils.tokenizer import (AnyTokenizer, MistralTokenizer,
get_cached_tokenizer)
from vllm.usage.usage_lib import UsageContext
from vllm.utils import Counter, Device, as_iter, is_list_of
from vllm.v1.engine.llm_engine import LLMEngine
from vllm.v1.sample.logits_processor import LogitsProcessor
if TYPE_CHECKING:
@@ -309,11 +308,7 @@ class LLM:
self.request_counter = Counter()
self.default_sampling_params: Union[dict[str, Any], None] = None
if envs.VLLM_USE_V1:
supported_tasks = self.llm_engine \
.get_supported_tasks() # type: ignore
else:
supported_tasks = self.llm_engine.model_config.supported_tasks
supported_tasks = self.llm_engine.get_supported_tasks() # type: ignore
logger.info("Supported_tasks: %s", supported_tasks)
@@ -1473,8 +1468,6 @@ class LLM:
Note:
This method is only available with the V1 LLM engine.
"""
from vllm.v1.engine.llm_engine import LLMEngine as V1LLMEngine
assert isinstance(self.llm_engine, V1LLMEngine)
return self.llm_engine.get_metrics()
def _validate_and_add_requests(