Add prompt_feedback/Safety filter info to VertexAI Gemini Response

[kardiff18]

Right now, a user receives an empty string of text is a Safety filter blocks the response rather than the information about the safety filters (such as the probability, etc.) It would be great to have the safety filter dictionary returned as part of prompt feedback, similar to the non-VertexAI Gemini implementation

langchain-google/libs/genai/langchain_google_genai/chat_models.py

Line 526 in c34ac8a

llm_output = {"prompt_feedback": proto.Message.to_dict(response.prompt_feedback)}

[lkuligin]

it's part of the generation_info:

langchain-google/libs/vertexai/langchain_google_vertexai/_utils.py

Line 151 in d4dcff3

"safety_ratings": [

or am I missing anything?

[kardiff18]

That's fair.. I guess i was hoping for the prompt_feedback directly. Based on the chain I'm using I can't get it from generation_info myself without writing a custom callback, which I can do it just would be nice if everything was returned natively.

You can close this for now.