Entities & syntax

The full parse: named entities highlighted inline, the dependency tree drawn as arcs, and every token's lemma, part-of-speech, fine-grained tag and head. Edit the text — it re-parses as you type.

Try typing into this. What do you think?

Parsere-parses live while you type

Word similarity

Cosine similarity between two words' 300-d vectors — straight from the embedding table baked into this file. The numbers are identical to spaCy's en_core_web_md: king/queen lands at 0.38 here exactly as it does there.

vsCompare

Text → embedding

Averages the token vectors into a single document embedding (doc.vector) — the vector you'd feed a clustering or semantic-search index. Each cell below is one of the 300 dimensions.

A dog chased the cat through the garden.

Embed

Lemmatizer

Reduces every token to its dictionary base form — running→run, better→well, mice→mouse — using the same rules and exception tables as spaCy.

The children were running faster and studying better than the mice.

Lemmatize