Set default similarity for Cohere model to cosine #125370
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes elastic#122878
|
Pinging @elastic/ml-core (Team:ML) |
|
Hi @jimczi, I've created a changelog YAML for you. |
davidkyle
approved these changes
Mar 21, 2025
There was a problem hiding this comment.
LGTM
There is a concern the a user might delete an Cohere inference endpoint with default similarity dot_product then recreate it but this time the similarity has defaulted to cosine. If that inference endpoint is used in semantic_text then this could cause suboptimal relevance.
This problem is described in #124272, that work should now be prioritised
We check the model configuration on every index request so this will result in an error. For users that deleted their endpoint and re-create it, they will need to override the similarity explicitly to match the old one. Since the user forced the deletion, I am not particularly concerned by this regression. WDYT? |
|
I see no reason not to merge this PR. It would be a better user experience to have the check performed up front and I will follow up on that |
jimczi
added a commit
to jimczi/elasticsearch
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes elastic#122878
jimczi
added a commit
to jimczi/elasticsearch
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes elastic#122878
smalyshev
pushed a commit
to smalyshev/elasticsearch
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes elastic#122878
elasticsearchmachine
pushed a commit
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes #122878
elasticsearchmachine
pushed a commit
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes #122878
elasticsearchmachine
pushed a commit
that referenced
this pull request
Mar 21, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes #122878
omricohenn
pushed a commit
to omricohenn/elasticsearch
that referenced
this pull request
Mar 28, 2025
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues,
our check ({@link DenseVectorFieldMapper#isNotUnitVector(float)}) often fails.
This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes elastic#122878
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Cohere embeddings are expected to be normalized to unit vectors, but due to floating point precision issues, our check (DenseVectorFieldMapper#isNotUnitVector) often fails. This change fixes this bug by setting the default similarity for newly created Cohere inference endpoint to cosine.
Closes #122878