You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Can you pass extra headers to the large language model and extract the headers in the response?
Motivation, pitch
I have AzureOpenAI end points in different regions and my use case requests that some of my data should not leave the country of origin. It's super easy to do that with AzureOpenAI chat completions, by passing in the target endpoint for each query, but I can't find the equivalent functionality with PandasAI Agents. Moreover, I don't not have any way of verifying which endpoint gave me my response, because I either get a string, dataframe or graph in the response.
I know it's possible because PandasAI sits above the language models which can accept and return extra response headers. I need to make a conversational data agent and PandasAI is ticking all the boxes - except that security one. Is this already possible and I missed it in the documentation somewhere?
Alternatives
No response
Additional context
No response
The text was updated successfully, but these errors were encountered:
🚀 The feature
Can you pass extra headers to the large language model and extract the headers in the response?
Motivation, pitch
I have AzureOpenAI end points in different regions and my use case requests that some of my data should not leave the country of origin. It's super easy to do that with AzureOpenAI chat completions, by passing in the target endpoint for each query, but I can't find the equivalent functionality with PandasAI Agents. Moreover, I don't not have any way of verifying which endpoint gave me my response, because I either get a string, dataframe or graph in the response.
I know it's possible because PandasAI sits above the language models which can accept and return extra response headers. I need to make a conversational data agent and PandasAI is ticking all the boxes - except that security one. Is this already possible and I missed it in the documentation somewhere?
Alternatives
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: