portkey = Portkey(
api_key="...",
config={
"override_params": {
"model": "gpt-4o"
},
"virtual_key": "openai-cxy-362aeb"
}
)Hello I am using the perplexity API. I have an open question query - i.e users are allowed to ask any question in an open search. I have the following system prompt: Your task is to respond to the question by analyzing the most recent content from the internet. Please put your response in a paragraph format and do not exceed more than 200 words. When answering do not cut off your sentences.. The responses get cut off - meaning the last sentence is usually incomplete and doesn't finish the thought. In the user prompt I simply have "{{question}}" . Any ideas on how to not let Perplexity provide sentences at the end that are cut off? Do I need to add more context in the User prompt in Portkey?
Also curious if anyone has figured out how to get the sources from Perplexity API to show up in the responses as it does it in web interface
"tool_choice": "required" in the SDK or the API as per OpenAI format {
"model": "gemini-1.5-pro",
"max_tokens": 200,
"stream": false,
"messages": [
{
"role": "system",
"content": "Say Hi"
},
{
"role": "user",
"content": "Hey, what is the temperature in San Francisco in Celsius? Also tell me the current time"
}
],
"tool_choice": "required",
"tools": [
....
]
}