Hey everyone,
How does he come to the conclusion that a lower value is better as for cosine similarity a bigger value is more similar?
Thnks, best Mike
Please give the video title and a time mark.
Video Name: Vector Database
Time: 4min 25 seconds
He’s using cosine distance, not cosine similarity.
From the Wikipedia article on cosine similarity:
Note that cosine distance is (1 - the cosine similarity).
1 Like
Ah, not listening properly enough. Thanks for clarification TMosh