[Frontend] Use new Renderer for Completions and Tokenize API (#32863)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-01-31 04:51:15 -08:00
committed by GitHub
parent 8980001c93
commit f0a1c8453a
64 changed files with 2116 additions and 2003 deletions
+27 -3
View File
@@ -3,11 +3,10 @@
import time
from typing import Any, Generic, TypeAlias, TypeVar
from pydantic import (
Field,
)
from pydantic import Field
from vllm import PoolingParams
from vllm.config import ModelConfig
from vllm.config.pooler import get_use_activation
from vllm.entrypoints.openai.engine.protocol import OpenAIBaseModel, UsageInfo
from vllm.entrypoints.pooling.base.protocol import (
@@ -18,6 +17,7 @@ from vllm.entrypoints.pooling.base.protocol import (
EncodingRequestMixin,
PoolingBasicRequestMixin,
)
from vllm.renderers import TokenizeParams
from vllm.tasks import PoolingTask
from vllm.utils import random_uuid
@@ -30,6 +30,18 @@ class PoolingCompletionRequest(
):
task: PoolingTask | None = None
def build_tok_params(self, model_config: ModelConfig) -> TokenizeParams:
encoder_config = model_config.encoder_config or {}
return TokenizeParams(
max_total_tokens=model_config.max_model_len,
max_output_tokens=0,
truncate_prompt_tokens=self.truncate_prompt_tokens,
do_lower_case=encoder_config.get("do_lower_case", False),
add_special_tokens=self.add_special_tokens,
max_total_tokens_param="max_model_len",
)
def to_pooling_params(self):
return PoolingParams(
truncate_prompt_tokens=self.truncate_prompt_tokens,
@@ -48,6 +60,18 @@ class PoolingChatRequest(
description=("Additional kwargs to pass to the HF processor."),
)
def build_tok_params(self, model_config: ModelConfig) -> TokenizeParams:
encoder_config = model_config.encoder_config or {}
return TokenizeParams(
max_total_tokens=model_config.max_model_len,
max_output_tokens=0,
truncate_prompt_tokens=self.truncate_prompt_tokens,
do_lower_case=encoder_config.get("do_lower_case", False),
add_special_tokens=self.add_special_tokens,
max_total_tokens_param="max_model_len",
)
def to_pooling_params(self):
return PoolingParams(
truncate_prompt_tokens=self.truncate_prompt_tokens,