Cosign Distance in an Embedded
Space
[DuckDuckGo's AI response when
it could not find a web reference to the term "cosign distance
in an embedded space":]
Cosine distance measures how different two vectors are by
looking at the angle between them, ignoring their size. Two key
ideas define it: cosine similarity and the distance math.
How It Works
- Angle
check: It looks at the direction of two vectors
in a space.
- Size
ignored: Long and short arrows pointing the
same way score the same.
- Cosine
similarity: Measures how close the vectors
point. A score of 1 means they point the exact same way. A
score of 0 means they are at a right angle (unrelated). A
score of -1 means they point opposite ways.
- Cosine
distance: Found by taking 1 and subtracting the
cosine similarity (1 - similarity).
- Value
range: Ranges from 0 (meaning identical
direction) to 2 (meaning opposite direction).
- Meaning
check: Words or sentences with similar meanings
get mapped close together.
- Finding
matches: AI tools use this to find text or
items that match your search.