How to Train Generative AI Using Company’s Data?

[liopolt09]

I am exploring the utilization of Vertex AI to create an AI-powered customer service solution capable of addressing complaints and providing responses to Q&A inquiries. The company I am associated with has generously supplied me with a significant amount of updated data that is not publicly available on the Internet. Employing Palm2 from the model garden, I have inputted textual information by incorporating it as a prompt. However, I've encountered a limitation issue with the following error message: 'Error during prediction: 3 INVALID_ARGUMENT: The request cannot be processed. The most likely reason is that the provided input exceeded the model's input token limit.'

I am seeking guidance on an effective approach to seamlessly integrate information about our company and its business into the AI chat's knowledge base. Your insights on a sound strategy for integrating such data into the AI's knowledge base would be highly valuable and appreciated.

[edudesouza]

Same here, I would love to know that also

[Renato_Nobre1]

If it is a pure Q&A why dont you use Vertex AI Conversation adding your company data as datastore (eg: csv)