Update specification

Author: Jan-Kazlouski-elastic

output/openapi/elasticsearch-openapi.json

@@ -1524,7 +1524,7 @@ export class LlamaServiceSettings {
   /**
    * The URL endpoint of the Llama stack endpoint.
    * URL must contain:
-   * * For `text_embedding` task - `/v1/inference/embeddings`.
+   * * For `text_embedding` task - `/v1/openai/v1/embeddings`.
    * * For `completion` and `chat_completion` tasks - `/v1/openai/v1/chat/completions`.
    */
   url: string
@@ -1552,7 +1552,9 @@ export class LlamaServiceSettings {
    */
   max_input_tokens?: integer
   /**
-   * For a `text_embedding` task, the number of dimensions the resulting output embeddings should have.
+   * For a `text_embedding` task, the number of dimensions the resulting output embeddings must have.
+   * It is supported only in `text-embedding-3` and later models. If it is not set by user, it defaults to the model returned dimensions.
+   * If model returns embeddings with a different number of dimensions, error is returned.
    */
   dimensions?: integer
   /**
@@ -1566,6 +1568,14 @@ export class LlamaServiceSettings {
   rate_limit?: RateLimitSetting
 }
 
+export class LlamaTaskSettings {
+  /**
+   * For a `completion` or `text_embedding` task, specify the user issuing the request.
+   * This information can be used for abuse detection.
+   */
+  user?: string
+}
+
 export enum LlamaTaskType {
   text_embedding,
   completion,

output/openapi/elasticsearch-serverless-openapi.json

@@ -23,6 +23,7 @@ import { Duration } from '@_types/Time'
 import {
   LlamaServiceSettings,
   LlamaServiceType,
+  LlamaTaskSettings,
   LlamaTaskType
 } from '@inference/_types/CommonTypes'
 import { InferenceChunkingSettings } from '@inference/_types/Services'
@@ -75,5 +76,10 @@ export interface Request extends RequestBase {
      * Settings used to install the inference model. These settings are specific to the `llama` service.
      */
     service_settings: LlamaServiceSettings
+    /**
+     * Settings to configure the inference task.
+     * These settings are specific to the task type you specified.
+     */
+    task_settings?: LlamaTaskSettings
   }
 }

output/schema/schema.json

@@ -6,7 +6,8 @@ value: |-
   {
     "service": "llama",
     "service_settings": {
-      "url": "http://localhost:8321/v1/inference/embeddings"
+      "url": "http://localhost:8321/v1/openai/v1/embeddings"
+      "dimensions": 384,
       "api_key": "llama-api-key",
       "model_id": "all-MiniLM-L6-v2" 
     }