[ML] Move to the Cohere V2 API for new inference endpoints #129884
+2,214
−1,360
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
davidkyle
added
>enhancement
:ml
|
Pinging @elastic/ml-core (Team:ML) |
|
Hi @davidkyle, I've created a changelog YAML for you. |
davidkyle
commented
Jun 23, 2025
| @@ -166,24 +166,14 @@ private static CohereModel createModel( | |||
| return switch (taskType) { | |||
| case TEXT_EMBEDDING -> new CohereEmbeddingsModel( | |||
| inferenceEntityId, | |||
| taskType, | |||
There was a problem hiding this comment.
Just removing the task type and service name parameters as both of these are known from model type.
| builder.endObject(); | ||
| return builder; | ||
| } | ||
|
|
||
| public XContentBuilder toXContentFragment(XContentBuilder builder, Params params) throws IOException { | ||
| return toXContentFragmentOfExposedFields(builder, params); | ||
| toXContentFragmentOfExposedFields(builder, params); | ||
| return builder.field(API_VERSION, apiVersion); // API version is persisted but not exposed to the user |
There was a problem hiding this comment.
Moving this to the toXContent will expose it to the user
| } | ||
|
|
||
| @Override | ||
| public HttpRequest createHttpRequest() { |
There was a problem hiding this comment.
Moving shared code to the base class
| import java.util.List; | ||
| import java.util.Objects; | ||
|
|
||
| public class CohereV1CompletionRequest extends CohereRequest { |
There was a problem hiding this comment.
The CohereCompletionRequest and CohereCompletionRequestEntity classes into a single class here. Same for the other request types, this is a refactoring not new code
|
|
||
| @Override | ||
| protected List<String> pathSegments() { | ||
| return List.of(CohereUtils.VERSION_2, CohereUtils.CHAT_PATH); |
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.
The Cohere V2 API contains 2 changes that must be adapted for
modelparameter is no longer optionalinput_typeparameter is no longer optionalCreating an endpoint without a model now causes a validation exception.
input_typeis declared either in task_settings or in the inference call, if not set in either or these places input_type defaults to search_query.New inference endpoints will use the V2 API, existing endpoints will continue to use the V1 API. The user does not have the option of picking the V1 API in new endpoints. One possibly controversial aspect is that the API version is not surfaced to the user, the version is persisted with the model config but not included in the
GET _inferenceresponse. I implemented this behaviour because the user does not have the ability to pick the API, in retrospect hiding the version now seems confusing.The request classes have been moved to
org.elasticsearch.xpack.inference.services.cohere.request.v1and renamed. The new V2 request classes are inorg.elasticsearch.xpack.inference.services.cohere.request.v2(they are very similar).The upgrade test
CohereServiceUpgradeITtests that the old v1 endpoints still work after upgrading.