-
Notifications
You must be signed in to change notification settings - Fork 260
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Arrow casting errors during/after KNN search when vectors are float16, if a vector index is created with gpu #3478
Comments
FYI @BubbleCal . I wonder if we are assuming the vectors are f32 somewhere here? Would be good to rerun this with |
Sorry I should have provided that myself. The full traceback is below. It looks like the relevant bit might be this bit?
Full traceback
|
Thanks! that traceback helps narrow it down. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Problem
The docs suggest that vectors can be written as
float16
to save disk space. However, doing this causes an error in arrow-array when I perform a KNN search withto_table(nearest=...)
. Strangely this only occurs if a vector index is built on gpu (see MWE below).Is this a bug or am I doing something wrong?
Relevant dependency verisons
MWE
Also see colab gist.
Result if
INDEX=False
orINDEX=True
andACCELERATOR = None
:Result if
INDEX=True
andACCELERATOR = 'cuda'
:The same problem occurs if the
FixedShapeTensorArray
is created asfloat32
and then casted tofloat16
as done in the docs:The text was updated successfully, but these errors were encountered: