-
Notifications
You must be signed in to change notification settings - Fork 12
Description
Is your feature request related to a problem?
Several benchmarks rely on datasets which were generated using dot product similarity function. When using innerproduct as a similarity measure, the code errors out due to:
failed shard on node [ZFEA9QwOTfC9bKAQ0skBCw]: shard failure, reason [lucene commit failed], failure IllegalArgumentException[Unsupported similarity function: MAXIMUM_INNER_PRODUCT]"
What solution would you like?
Add support for DotProduct in the jvector plugin layer.
What alternatives have you considered?
A clear and concise description of any alternative solutions or features you've considered.
Do you have any additional context?
When considering this solution we'd need to normalize the vector values, which is required by Lucene.