- 🪜The chain works in two steps:
- 1️⃣First, it rephrases the input question into a "standalone" question, dereferencing pronouns based on the chat history.
- 2️⃣Then, it queries the retriever for documents similar to the dereferenced question and composes an answer.
- 🐶By default, the agent is pretending to be a talking puppy, but you can change the prompt to whatever you want!
- 👇Upload some text, then try asking e.g.
What is a document loader?below!