Buster, the QA documentation chatbot!
Buster is a question-answering chatbot that can be tuned to specific documentations. You can try it here, where it will answer questions about 🤗 Transformers.
How does Buster work?
First, we parsed the documentation into snippets. For each snippet, we obtain an embedding by using the OpenAI API.
Then, when a user asks a question, we compute its embedding, and find the snippets from the doc with the highest cosine similarity to the question.
Finally, we craft the prompt:
- The most relevant snippets from the doc.
- The engineering prompt.
- The user's question.
We send the prompt to the OpenAI API, and display the answer to the user!
Currently used models
- For embeddings: "text-embedding-ada-002"
- For completion: "text-davinci-003"
Livestream
For more information, you can watch the livestream where explain how buster works in detail!