Poor-man's oss perplexity.ai/phind.com implementation
Originally for AGI Hackathon on Vector Search 01/27, see hackathon-01-27 branch for the version built in hackathon.
- The UI is adapted from https://github.com/leptonai/search_with_lepton.
- Language model inference is implemented using an OpenAI-compatible interface.
- A vector embedding-based cache is implemented for query embeddings; vector search is conducted via convex.dev.