[BugFix][Frontend] Use LoRA tokenizer in OpenAI APIs (#6227)

Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
This commit is contained in:
Nick Hill
2024-07-18 15:13:30 +08:00
committed by GitHub
co-authored by Cyrus Leung
parent 8a74c68bd1
commit e2fbaee725
16 changed files with 267 additions and 186 deletions
+4 -7
View File
@@ -89,14 +89,11 @@ class OpenAIServingEmbedding(OpenAIServing):
prompt_is_tokens, prompts = parse_prompt_format(request.input)
pooling_params = request.to_pooling_params()
tokenizer = await self.engine.get_tokenizer()
for i, prompt in enumerate(prompts):
if prompt_is_tokens:
prompt_formats = self._validate_prompt_and_tokenize(
request, prompt_ids=prompt)
else:
prompt_formats = self._validate_prompt_and_tokenize(
request, prompt=prompt)
prompt_arg = "prompt_ids" if prompt_is_tokens else "prompt"
prompt_formats = await self._validate_prompt_and_tokenize(
request, tokenizer, **{prompt_arg: prompt})
prompt_ids, prompt_text = prompt_formats
generator = self.engine.encode(