Xenova/all-MiniLM-L6-v2
Semantic similarity
First run: approximately 25 MBQuantized weights plus tokenizer and configuration files.
Nothing downloads until you press the button.
LAB 002 · Embeddings
Compare meaning, locally.
Turn two English passages into normalized 384-dimensional vectors, then calculate their similarity on your browser CPU.
First download≈25 MB
- Runtime
- WebAssembly
- Input uploaded
- None
- API key
- None
CPU · local
Tokenize
Each passage is converted into tokens understood by the model.
Embed
The model produces a normalized 384-value representation for each passage.
Compare
A dot product measures vector alignment, from different to highly similar.