panyy
/
UnisKB
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: Vllm url support v1

v3.2
zhangzhanwei 2025-09-09 13:46:32 +08:00 committed by zhanweizhang7
parent 6ddad88357
commit d2463abaac
3 changed files with 10 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
apps/chat/serializers/chat.py
View File

@ -158,9 +158,12 @@ class PromptGenerateSerializer(serializers.Serializer):
q = prompt.replace("{userInput}", message) q = prompt.replace("{userInput}", message)
messages[-1]['content'] = q messages[-1]['content'] = q
model_exist = QuerySet(Model).filter(workspace_id=workspace_id, id=model_id).exists() model_exist = QuerySet(Model).filter(workspace_id=workspace_id,
id=model_id,
model_type = "LLM"
).exists()
if not model_exist: if not model_exist:
raise Exception(_("model does not exists")) raise Exception(_("Model does not exists or is not an LLM model"))
def process(): def process():
model = get_model_instance_by_model_workspace_id(model_id=model_id, workspace_id=workspace_id) model = get_model_instance_by_model_workspace_id(model_id=model_id, workspace_id=workspace_id)

3
apps/models_provider/impl/vllm_model_provider/model/reranker.py
View File

@ -28,10 +28,11 @@ class VllmBgeReranker(MaxKBBaseModel, BaseDocumentCompressor):
@staticmethod @staticmethod
def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs): def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
r_url = model_credential.get('api_url')[:-3] if model_credential.get('api_url').endswith('/v1') else model_credential.get('api_url')
return VllmBgeReranker( return VllmBgeReranker(
model=model_name, model=model_name,
api_key=model_credential.get('api_key'), api_key=model_credential.get('api_key'),
api_url=model_credential.get('api_url'), api_url=r_url,
params=model_kwargs, params=model_kwargs,
**model_kwargs **model_kwargs
) )

4
apps/models_provider/impl/vllm_model_provider/model/whisper_sst.py
View File

@ -44,7 +44,9 @@ class VllmWhisperSpeechToText(MaxKBBaseModel, BaseSpeechToText):
self.speech_to_text(audio_file) self.speech_to_text(audio_file)
def speech_to_text(self, audio_file): def speech_to_text(self, audio_file):
base_url = f"{self.api_url}/v1"
base_url = self.api_url if self.api_url.endswith('v1') else f"{self.api_url}/v1"
try: try:
client = OpenAI( client = OpenAI(
api_key=self.api_key, api_key=self.api_key,