BPE Tokenization Visualizer

Speed:
1x
Step 1 of 0

BPE Steps

Tokenization Visualization

Token Pair Frequencies

Pair Frequencies

What is this?
No token pairs available

The BPE algorithm selects the most frequent pair (highlighted in yellow) to merge in each step. This creates a new token that replaces all occurrences of that pair.

Vocabulary Statistics

Vocabulary Size
0
unique tokens
Compression Ratio
1.00
original length / current length

Current Vocabulary