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

[Frontend][Core] Move guided decoding params into sampling params (#8252)

Browse Source 
Signed-off-by: Joe Runde <Joseph.Runde@ibm.com>
Co-authored-by: Nick Hill <nickhill@us.ibm.com>
This commit is contained in:
Joe Runde
2024-09-30 19:34:25 -06:00
committed by GitHub
parent bce324487a
commit 062c89e7c9
16 changed files with 441 additions and 281 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
vllm/entrypoints/openai/serving_engine.py
View File

@@ -27,11 +27,9 @@ from vllm.entrypoints.openai.protocol import (ChatCompletionRequest,
from vllm.inputs.parse import parse_and_batch_prompt
from vllm.logger import init_logger
from vllm.lora.request import LoRARequest
from vllm.model_executor.guided_decoding import (
get_guided_decoding_logits_processor)
from vllm.pooling_params import PoolingParams
from vllm.prompt_adapter.request import PromptAdapterRequest
from vllm.sampling_params import LogitsProcessor, SamplingParams
from vllm.sampling_params import SamplingParams
from vllm.sequence import Logprob
from vllm.transformers_utils.tokenizer import AnyTokenizer
from vllm.utils import AtomicCounter
@@ -168,15 +166,6 @@ class OpenAIServing:
})
return json_str
async def _guided_decode_logits_processor(
self, request: Union[ChatCompletionRequest, CompletionRequest],
tokenizer: AnyTokenizer) -> Optional[LogitsProcessor]:
decoding_config = await self.engine_client.get_decoding_config()
guided_decoding_backend = request.guided_decoding_backend \
or decoding_config.guided_decoding_backend
return await get_guided_decoding_logits_processor(
guided_decoding_backend, request, tokenizer)
async def _check_model(
self,
request: AnyRequest,