If you need to quickly build a prototype using a customized LLM flow, you can use this nice open UI visual tool called Flowise and it’s open source. It offers a builin templates that .
Requirements for this:
After cloning, installing and running Flowise you will see an interface like this:
Head to Marketplaces and choose Conversational Retrieval QA Chain template. Then save as Chatbot and get ready to paste your OpenAI keys into the 2 modals and Pinecone API key as well with it’s environment “us-west4-gcp”.
You need to create an index in Pinecone with a name and importantly the dimensions of 1536, number that OpenAI uses with their embeddings.
For this template you can upload a text file with some context about your product or project, like a README file. I decided to replace the document loader input (Text File) for a Github repository (https://github.com/openai/whisper) and reconnected the text and document inputs. That’s it. You can expand the chatbot in the top right side and start interacting with your model.
How to setup please?