Skip to content

[EIS] Rename the elser 2 default model and the default inference endpoint #130336

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

Merged

Conversation

timgrein
Copy link
Contributor

We've decided to rename the default model and default inference endpoint for elser running EIS. This needs to be backported to 9.1 and 8.19 and we consider it as a bug as multiple other places (will) depend on these new names

@timgrein timgrein added >bug :ml Machine learning Team:ML Meta label for the ML team v8.19.0 v9.1.0 labels Jun 30, 2025
@elasticsearchmachine
Copy link
Collaborator

Pinging @elastic/ml-core (Team:ML)

@elasticsearchmachine
Copy link
Collaborator

Hi @timgrein, I've created a changelog YAML for you.

@@ -41,7 +41,7 @@ public void testGetDefaultEndpoints() throws IOException {
}

assertInferenceIdTaskType(allModels, ".rainbow-sprinkles-elastic", TaskType.CHAT_COMPLETION);
assertInferenceIdTaskType(allModels, ".elser-v2-elastic", TaskType.SPARSE_EMBEDDING);
assertInferenceIdTaskType(allModels, ".elser-2-elastic", TaskType.SPARSE_EMBEDDING);
assertInferenceIdTaskType(allModels, ".multilingual-embed-v1-elastic", TaskType.TEXT_EMBEDDING);
assertInferenceIdTaskType(allModels, ".rerank-v1-elastic", TaskType.RERANK);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need to remove v from the text embedding and rerank default endpoint names too?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nope, we'll only do the adjustment for ELSER to be en par with the ML node

Copy link
Contributor

@brendan-jugan-elastic brendan-jugan-elastic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! One quick question though to help my understanding

static final String DEFAULT_ELSER_ENDPOINT_ID_V2 = defaultEndpointId(DEFAULT_ELSER_MODEL_ID_V2);
// elser-2
static final String DEFAULT_ELSER_2_MODEL_ID = "elser_model_2";
static final String DEFAULT_ELSER_ENDPOINT_ID_V2 = defaultEndpointId("elser-2");
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't love this mismatch, but I imagine there are some considerations you're keeping in mind. Does changing it have backwards compatibility ramifications? Trying to think what the impact is, but can't figure it out 😅

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm with you, it doesn't look great. But the goal here is to be consistent with how things are called in the ML node, at least for ELSER. The model is called .elser_model_2 and the inference endpoint is .elser-2-elasticsearch. If we expect people to migrate with less confusion, it's good if the names are similar.

Now, the name of the reranker is .rerank-v1. That's how we got here...

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

backwards compatibility

These endpoints are not released yet, so we have the luxury of being able to change the name still.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ahh, gotcha. Thanks for the clarification!

@timgrein timgrein merged commit c1b43d8 into elastic:main Jul 3, 2025
33 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
>bug :ml Machine learning Team:ML Meta label for the ML team v8.19.0 v9.1.0 v9.2.0
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants