Activity v1
RAG Pipeline and Retrieval Parameter Visualizer
Tune chunk size retrieval breadth hybrid weight and reranking in a deterministic pipeline.
Retrieval trade-off model
Illustrative equations expose directional trade-offs only. They are not benchmark predictions and must be replaced by corpus-specific evaluation.
Estimated recall91%
Estimated precision89%
Context tokens1,600
Parse + 400 token chunks
→Hybrid retrieve k=8
→Rerank top set
→Grounded context
Expected outcomes
- Relate parameters to precision recall and cost
- Identify retrieval bottlenecks