Activity v1

RAG Pipeline and Retrieval Parameter Visualizer

Tune chunk size retrieval breadth hybrid weight and reranking in a deterministic pipeline.

completion
beginner
seed 11

RAG Pipeline and Retrieval Parameter Visualizer: Baseline

Explore the core controls with stable inputs and visible assumptions.

Assumptions

The simulation is deterministic and intentionally simplifies provider and hardware behavior.

Failure injection

Stable baseline with no injected production fault.

RAG Chunking & Vector Search Visualizer

Interactive playground for text segmentation, distance metric calculators, and HNSW vector graph traversal.

6 CHUNKS GENERATED
~194 TOTAL TOKENS
Chunking Strategy & Parameters
Chunk Size180 chars
Overlap Window40 chars
Input Document & Chunk Boundary Highlighting814 characters
Generated Passages (Click chunk to inspect vector details):
CHUNK #1 INSPECTOR
180 chars · ~43 tokens
Cosine Similarity:89.7%
"Retrieval-Augmented Generation (RAG) grounds Large Language Model responses in verified domain knowledge. Instead of relying solely on parametric weights learned during pre-trainin"
Dense Vector Embeddings (4D Space)
Semantic Density
0.913
Domain Relevance
0.667
Lexical Match
0.312
Context Depth
0.285
Cosine Similarity Formula
cos(θ) = (A · B) / (||A|| × ||B||) = 0.897

Measures directional alignment independent of vector magnitude. Values range from 0.0 to 1.0.

HNSW Vector Graph Traversal Stepper
Layer 2: Express Entry PointActive: Node_L2_Top (Entry)

Search begins at sparse top layer. Performs greedy distance routing across long-distance highway edges.

Candidate set: [Node_L2_Top] (Distance = 0.82)
Entry (Layer 2)Coarse Hop (Layer 1)Top-1 Nearest NeighborCandidate Block
0 saved attempts

Expected outcomes

  • Relate parameters to precision recall and cost
  • Identify retrieval bottlenecks

Connected concepts

Apply this lab in an architecture