AI demo — sentence meaning as vectors
Type a sentence. The page turns words into embeddings, projects them to 1D (PCA), and draws a “wave” (x = word position, y = projected meaning). Below, you’ll also see semantic similarity (cosine) between sentences and “closest words”.
Model:
all-MiniLM-L6-v2
Insights
Similarity uses cosine distance on sentence embeddings (mean of word embeddings).
The wave is a 1D projection (PCA) fitted on the words currently in memory.
0
sentences
0
unique words
History (this tab only)
Each added sentence is listed with a mini-wave. Reloading the page clears everything.