ROADMAP.md

Next Version

• Add DSpy for prompt-tuning to make small models(Qwen2 7B, Llama 3.1 8B...) can extract entities. @NumberChiffre @gusye1234
• Optimize Algorithm: add global_local query method, globally rewrite query then perform local search.

In next few versions

• Add rate limiter: support token limit (tokens per second, per minute)

• Add other advanced RAG algorithms, candidates:

  • HybridRAG
  • HippoRAG

Interesting directions

• Add Sciphi Triplex as the entity extraction model.
• Add new components, see issue