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

[Frontend] Add tokenize/detokenize endpoints (#5054)

Browse Source
This commit is contained in:
sasha0552
2024-06-26 16:54:22 +00:00
committed by GitHub
parent 5bfd1bbc98
commit c54269d967
5 changed files with 143 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
vllm/entrypoints/openai/serving_completion.py
View File

@@ -16,7 +16,11 @@ from vllm.entrypoints.openai.protocol import (CompletionLogProbs,
CompletionResponseChoice,
CompletionResponseStreamChoice,
CompletionStreamResponse,
UsageInfo)
DetokenizeRequest,
DetokenizeResponse,
TokenizeRequest,
TokenizeResponse, UsageInfo)
# yapf: enable
from vllm.entrypoints.openai.serving_engine import (LoRAModulePath,
OpenAIServing)
from vllm.logger import init_logger
@@ -442,3 +446,29 @@ class OpenAIServingCompletion(OpenAIServing):
tokens=out_tokens,
top_logprobs=out_top_logprobs,
)
async def create_tokenize(self,
request: TokenizeRequest) -> TokenizeResponse:
error_check_ret = await self._check_model(request)
if error_check_ret is not None:
return error_check_ret
(input_ids, input_text) = self._validate_prompt_and_tokenize(
request,
prompt=request.prompt,
add_special_tokens=request.add_special_tokens)
return TokenizeResponse(tokens=input_ids,
count=len(input_ids),
max_model_len=self.max_model_len)
async def create_detokenize(
self, request: DetokenizeRequest) -> DetokenizeResponse:
error_check_ret = await self._check_model(request)
if error_check_ret is not None:
return error_check_ret
(input_ids, input_text) = self._validate_prompt_and_tokenize(
request, prompt_ids=request.tokens)
return DetokenizeResponse(prompt=input_text)