wang.yuqiandCyrus Leung
eeb549a74d
Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io> Signed-off-by: wang.yuqi <noooop@126.com> Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
64 lines
2.1 KiB
Python
64 lines
2.1 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
from collections.abc import Iterable, Sequence
|
|
from typing import TYPE_CHECKING
|
|
|
|
from transformers import PreTrainedTokenizerBase
|
|
|
|
from vllm.entrypoints.generate.base.protocol import DeltaMessage
|
|
from vllm.reasoning import ReasoningParser
|
|
|
|
if TYPE_CHECKING:
|
|
from vllm.entrypoints.openai.chat_completion.protocol import ChatCompletionRequest
|
|
from vllm.entrypoints.openai.responses.protocol import ResponsesRequest
|
|
|
|
|
|
class GptOssReasoningParser(ReasoningParser):
|
|
"""
|
|
Reasoning parser for GptOss model.
|
|
|
|
The GptOss model uses harmony to extract reasoning content and this parser
|
|
is only used for detecting the end of the reasoning content.
|
|
"""
|
|
|
|
def __init__(self, tokenizer: PreTrainedTokenizerBase, *args, **kwargs):
|
|
super().__init__(tokenizer, *args, **kwargs)
|
|
|
|
def is_reasoning_end(self, input_ids: Sequence[int]) -> bool:
|
|
return True
|
|
|
|
def is_reasoning_end_streaming(
|
|
self, input_ids: Sequence[int], delta_ids: Iterable[int]
|
|
) -> bool:
|
|
return True
|
|
|
|
def extract_content_ids(self, input_ids: list[int]) -> list[int]:
|
|
raise NotImplementedError(
|
|
"GptOssReasoningParser only provides boundary detection. "
|
|
"Use HarmonyParser for output parsing."
|
|
)
|
|
|
|
def extract_reasoning_streaming(
|
|
self,
|
|
previous_text: str,
|
|
current_text: str,
|
|
delta_text: str,
|
|
previous_token_ids: Sequence[int],
|
|
current_token_ids: Sequence[int],
|
|
delta_token_ids: Sequence[int],
|
|
) -> DeltaMessage | None:
|
|
raise NotImplementedError(
|
|
"GptOssReasoningParser only provides boundary detection. "
|
|
"Use HarmonyParser for output parsing."
|
|
)
|
|
|
|
def extract_reasoning(
|
|
self,
|
|
model_output: str,
|
|
request: "ChatCompletionRequest | ResponsesRequest",
|
|
) -> tuple[str | None, str | None]:
|
|
raise NotImplementedError(
|
|
"GptOssReasoningParser only provides boundary detection. "
|
|
"Use HarmonyParser for output parsing."
|
|
)
|