phoenix/litellm
1
0
Fork 0
forked from phoenix/litellm-mirror
Code Pull requests Activity

(feat) vertex ai - better debugging

Browse source
This commit is contained in:
ishaan-jaff 2023-12-07 09:38:14 -08:00
parent e9d93c624c
commit d2eee342fb
1 changed files with 16 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

19
litellm/llms/vertex_ai.py
View file

@ -93,45 +93,58 @@ def completion(
prompt = " ".join([message["content"] for message in messages]) prompt = " ".join([message["content"] for message in messages])
mode = "" mode = ""
request_str = ""
if model in litellm.vertex_chat_models: if model in litellm.vertex_chat_models:
chat_model = ChatModel.from_pretrained(model) chat_model = ChatModel.from_pretrained(model)
mode = "chat" mode = "chat"
request_str += f"chat_model = ChatModel.from_pretrained({model})\n"
elif model in litellm.vertex_text_models: elif model in litellm.vertex_text_models:
text_model = TextGenerationModel.from_pretrained(model) text_model = TextGenerationModel.from_pretrained(model)
mode = "text" mode = "text"
request_str += f"text_model = TextGenerationModel.from_pretrained({model})\n"
elif model in litellm.vertex_code_text_models: elif model in litellm.vertex_code_text_models:
text_model = CodeGenerationModel.from_pretrained(model) text_model = CodeGenerationModel.from_pretrained(model)
mode = "text" mode = "text"
request_str += f"text_model = CodeGenerationModel.from_pretrained({model})\n"
else: # vertex_code_chat_models else: # vertex_code_chat_models
chat_model = CodeChatModel.from_pretrained(model) chat_model = CodeChatModel.from_pretrained(model)
mode = "chat" mode = "chat"
request_str += f"chat_model = CodeChatModel.from_pretrained({model})\n"
if mode == "chat": if mode == "chat":
chat = chat_model.start_chat() chat = chat_model.start_chat()
request_str+= f"chat = chat_model.start_chat()\n"
## LOGGING ## LOGGING
logging_obj.pre_call(input=prompt, api_key=None, additional_args={"complete_input_dict": optional_params})
if "stream" in optional_params and optional_params["stream"] == True: if "stream" in optional_params and optional_params["stream"] == True:
# NOTE: VertexAI does not accept stream=True as a param and raises an error, # NOTE: VertexAI does not accept stream=True as a param and raises an error,
# we handle this by removing 'stream' from optional params and sending the request # we handle this by removing 'stream' from optional params and sending the request
# after we get the response we add optional_params["stream"] = True, since main.py needs to know it's a streaming response to then transform it for the OpenAI format # after we get the response we add optional_params["stream"] = True, since main.py needs to know it's a streaming response to then transform it for the OpenAI format
optional_params.pop("stream", None) # vertex ai raises an error when passing stream in optional params optional_params.pop("stream", None) # vertex ai raises an error when passing stream in optional params
request_str += f"chat.send_message_streaming({prompt}, **{optional_params})\n"
logging_obj.pre_call(input=prompt, api_key=None, additional_args={"complete_input_dict": optional_params, "request_str": request_str})
model_response = chat.send_message_streaming(prompt, **optional_params) model_response = chat.send_message_streaming(prompt, **optional_params)
optional_params["stream"] = True optional_params["stream"] = True
return model_response return model_response
request_str += f"chat.send_message({prompt}, **{optional_params}).text\n"
logging_obj.pre_call(input=prompt, api_key=None, additional_args={"complete_input_dict": optional_params, "request_str": request_str})
completion_response = chat.send_message(prompt, **optional_params).text completion_response = chat.send_message(prompt, **optional_params).text
elif mode == "text": elif mode == "text":
## LOGGING
logging_obj.pre_call(input=prompt, api_key=None)
if "stream" in optional_params and optional_params["stream"] == True: if "stream" in optional_params and optional_params["stream"] == True:
optional_params.pop("stream", None) # See note above on handling streaming for vertex ai optional_params.pop("stream", None) # See note above on handling streaming for vertex ai
request_str += f"text_model.predict_streaming({prompt}, **{optional_params})\n"
logging_obj.pre_call(input=prompt, api_key=None, additional_args={"complete_input_dict": optional_params, "request_str": request_str})
model_response = text_model.predict_streaming(prompt, **optional_params) model_response = text_model.predict_streaming(prompt, **optional_params)
optional_params["stream"] = True optional_params["stream"] = True
return model_response return model_response
request_str += f"text_model.predict({prompt}, **{optional_params}).text\n"
logging_obj.pre_call(input=prompt, api_key=None, additional_args={"complete_input_dict": optional_params, "request_str": request_str})
completion_response = text_model.predict(prompt, **optional_params).text completion_response = text_model.predict(prompt, **optional_params).text
## LOGGING ## LOGGING