Embedding Similarity Playground

See pairwise cosine similarity scores between sentences as a color-coded heatmap. Use pre-loaded examples or type your own sentences to understand how vector similarity drives semantic search.

Switch tool:Text Chunk Visualizer RAG Pipeline Explorer Token Counter & Visualizer Embedding Similarity Playground

Tunable Embedding Variables & Sample Sets

Choose pre-calibrated semantic clusters or tune similarity cutoff

Pre-Loaded Semantic Corpora

Similarity Cutoff Threshold (τ)Show All

0.00 (All pairs)0.50 (Moderate)0.90 (Near identical)

Sentences (5/10)

127c

229c

326c

426c

522c

Every input contains default text that you can freely edit or replace.

How Vector Similarity Works

Each sentence is transformed into a multi-dimensional embedding vector. The cosine similarity calculates the angle between these vectors: 1.0 means identical direction in semantic space, while 0.0 means completely uncorrelated vocabulary.

In production, neural embedding models like OpenAI text-embedding-3-small (1536d) capture subtle nuances, synonyms, and context beyond exact word matches.

Pairwise Cosine Matrix

5 × 5 Grid

Tap or hover any cell to inspect pair

Cosine Scale:

0.0 (Unrelated)

0.5

1.0 (Identical)

Pairwise Inspection: Sentence #1 vs Sentence #2

Cosine: 0.3651

Moderate Correlation

0.00 (Orthogonal)Angle: ~68.6°1.00 (Parallel)

Sentence #1

“How do I reset my password?”

Sentence #2

“I forgot my login credentials”

5 sentences

Semantic Search

SmartHub AI Engineering Track

Master AI Engineering with Production Projects

Go beyond visualizers — construct production-grade RAG pipelines, evaluate hallucination benchmarks, and deploy real AI applications alongside industry mentors.

Explore the Curriculum