TPM

Transformer Pattern Matcher & Visualizer

Query-Key Dot Products, Softmax Density & Context Gap Hallucination Mechanics
Pattern Matching Mechanics: High vector proximity between Query (prompt target) and Key vectors generates sharp softmax attention weights, enabling deterministic pattern deduction.
2D Token Embedding Space Drag Q/K vector tokens
Q-K Softmax Heatmap Normalized Weights

Next-Token Logit Distribution & Completion Stream

Calculated probability distribution based on softmax temperature scaling.
Sequence Continuation Output:
Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.