Embedding Visualizer
Downloading model... (slow on first visit)
Processes up to 8 words.
cosine similarity
0.00
Vector preview (first 12 dims)
No vectors yet.
Similarity matrix
No matrix yet.
3D visualization (PCA projection)
The 384-dimensional embeddings are projected onto the three axes of greatest variance (PC1·PC2·PC3). Drag to rotate, scroll to zoom. Notice how words in the same category (animals · fruits · IT devices) cluster together. Words you add yourself appear in gray.
💡Note: Notice how animals, fruits, and IT devices naturally form clusters in 3D space without anyone labeling them as such. Embeddings learn semantic structure on their own — purely from the contexts in which words appear, with no need for someone to teach "this is an animal".
Remixed by Bora Lee · Based on the original Tiktokenizer