how it works
every entry turns into a vector and your browser averages the vectors you picked into one picked vector, or one you passed into one passed vector. then we score it
manicule.pyrank()same math in rank.js
score=cos(entry, picked)what you picked
−λ · cos(entry, passed)what you passed × λ
two averages and a subtraction
the listingon github
manicule.py · rank()loading
one row, worked
- entry
- loading
the dashed part is what λ took
the same row, other λ
| λ | score | rank | what it does |
|---|
where it runs
- browser
- rank.js, the same math. corpus.json is 270 × 384 int8
- cli
- manicule.py rank. a markdown page from an opml and a folder of notes
- vectors
- fastembed, bge-small, on your machine
- state
- the url hash. #m=…
- refresh
- a github action, daily, 06:17 utc