Skip to content

alexanderkjackson-protonmail-com/ragman

Repository files navigation

ragman

Retrieval augmented generation for manual pages

Todo

  • Embeddings are simply every 512 tokens. Ideally CLS tokens should be generated for individual statements with the maximum available context.
  • Actually load everything into a Milvus collection.
  • Search functionality
  • Permit using alternate models
  • PyTorch (or alternative) interface for alternative models (for either preprocessing or embedding itself).

About

Retrieval augmented generation for manual pages

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages