feat(model): add model async trigger (#293)

Because - we are going to support asynchronous call for model trigger This commit - add model trigger async methods --------- Co-authored-by: droplet-bot <[email protected]>
instill-ai · Apr 2, 2024 · 3dcf05f · 3dcf05f
1 parent 3c32980
commit 3dcf05f
Show file tree

Hide file tree

Showing 3 changed files with 208 additions and 0 deletions.
diff --git a/model/model/v1alpha/model.proto b/model/model/v1alpha/model.proto
@@ -565,6 +565,32 @@ message TriggerUserModelResponse {
   repeated TaskOutput task_outputs = 2;
 }
 
+// TriggerAsyncUserModelRequest represents a request to trigger a model inference
+// asynchronously.
+message TriggerAsyncUserModelRequest {
+  // The resource name of the model , which allows its access by parent user
+  // and ID.
+  // - Format: `users/{user.id}/models/{model.id}`.
+  string name = 1 [
+    (google.api.field_behavior) = REQUIRED,
+    (google.api.resource_reference).type = "api.instill.tech/Model",
+    (grpc.gateway.protoc_gen_openapiv2.options.openapiv2_field) = {
+      field_configuration: {path_param_name: "user_model_name"}
+    }
+  ];
+  // Inference input parameters.
+  repeated TaskInput task_inputs = 2 [(google.api.field_behavior) = REQUIRED];
+  // Model version tag
+  string version = 3 [(google.api.field_behavior) = REQUIRED];
+}
+
+// TriggerAsyncUserModelResponse contains the information to access the
+// status of an asynchronous model inference.
+message TriggerAsyncUserModelResponse {
+  // Long-running operation information.
+  google.longrunning.Operation operation = 1 [(google.api.field_behavior) = OUTPUT_ONLY];
+}
+
 // TriggerUserModelBinaryFileUploadRequest represents a request trigger a model
 // inference by uploading a binary file as the input.
 message TriggerUserModelBinaryFileUploadRequest {
@@ -849,6 +875,32 @@ message TriggerOrganizationModelResponse {
   repeated TaskOutput task_outputs = 2;
 }
 
+// TriggerAsyncOrganizationModelRequest represents a request to trigger a model inference
+// asynchronously
+message TriggerAsyncOrganizationModelRequest {
+  // The resource name of the model , which allows its access by parent organization
+  // and ID.
+  // - Format: `organizations/{organization.id}/models/{model.id}`.
+  string name = 1 [
+    (google.api.field_behavior) = REQUIRED,
+    (google.api.resource_reference).type = "api.instill.tech/Model",
+    (grpc.gateway.protoc_gen_openapiv2.options.openapiv2_field) = {
+      field_configuration: {path_param_name: "organization_model_name"}
+    }
+  ];
+  // Inference input parameters.
+  repeated TaskInput task_inputs = 2 [(google.api.field_behavior) = REQUIRED];
+  // Model version tag
+  string version = 3 [(google.api.field_behavior) = REQUIRED];
+}
+
+// TriggerAsyncOrganizationModelResponse contains the information to access the
+// status of an asynchronous model inference.
+message TriggerAsyncOrganizationModelResponse {
+  // Long-running operation information.
+  google.longrunning.Operation operation = 1 [(google.api.field_behavior) = OUTPUT_ONLY];
+}
+
 // TriggerOrganizationModelBinaryFileUploadRequest represents a request trigger a model
 // inference by uploading a binary file as the input.
 message TriggerOrganizationModelBinaryFileUploadRequest {

diff --git a/model/model/v1alpha/model_public_service.proto b/model/model/v1alpha/model_public_service.proto
@@ -199,6 +199,18 @@ service ModelPublicService {
     option (google.api.method_signature) = "name,inputs";
   }
 
+  // Trigger model inference asynchronously
+  //
+  // Triggers a deployed model to infer the result of a set of task or
+  // questions.
+  rpc TriggerAsyncUserModel(TriggerAsyncUserModelRequest) returns (TriggerAsyncUserModelResponse) {
+    option (google.api.http) = {
+      post: "/v1alpha/{name=users/*/models/*}/{version=*}/triggerAsync"
+      body: "*"
+    };
+    option (google.api.method_signature) = "name,inputs";
+  }
+
   // Trigger model inference with a binary input
   //
   // Triggers a deployed model to infer the result of a task or question,
@@ -333,6 +345,18 @@ service ModelPublicService {
     option (google.api.method_signature) = "name,inputs";
   }
 
+  // Trigger model inference asynchronously
+  //
+  // Triggers a deployed model to infer the result of a set of task or
+  // questions.
+  rpc TriggerAsyncOrganizationModel(TriggerAsyncOrganizationModelRequest) returns (TriggerAsyncOrganizationModelResponse) {
+    option (google.api.http) = {
+      post: "/v1alpha/{name=organizations/*/models/*}/{version=*}/triggerAsync"
+      body: "*"
+    };
+    option (google.api.method_signature) = "name,inputs";
+  }
+
   // Trigger model inference with a binary input
   //
   // Triggers a deployed model to infer the result of a task or question,

diff --git a/openapiv2/model/service.swagger.yaml b/openapiv2/model/service.swagger.yaml
@@ -706,6 +706,48 @@ paths:
             $ref: '#/definitions/ModelPublicServiceTriggerUserModelBody'
       tags:
         - ModelPublicService
+  /v1alpha/{user_model_name}/{version}/triggerAsync:
+    post:
+      summary: Trigger model inference asynchronously
+      description: |-
+        Triggers a deployed model to infer the result of a set of task or
+        questions.
+      operationId: ModelPublicService_TriggerAsyncUserModel
+      responses:
+        "200":
+          description: A successful response.
+          schema:
+            $ref: '#/definitions/v1alphaTriggerAsyncUserModelResponse'
+        "401":
+          description: Returned when the client credentials are not valid.
+          schema: {}
+        default:
+          description: An unexpected error response.
+          schema:
+            $ref: '#/definitions/googlerpcStatus'
+      parameters:
+        - name: user_model_name
+          description: |-
+            The resource name of the model , which allows its access by parent user
+            and ID.
+            - Format: `users/{user.id}/models/{model.id}`.
+          in: path
+          required: true
+          type: string
+          pattern: users/[^/]+/models/[^/]+
+        - name: version
+          description: Model version tag
+          in: path
+          required: true
+          type: string
+          pattern: '[^/]+'
+        - name: body
+          in: body
+          required: true
+          schema:
+            $ref: '#/definitions/ModelPublicServiceTriggerAsyncUserModelBody'
+      tags:
+        - ModelPublicService
   /v1alpha/{organization_name}/models:
     get:
       summary: List organization models
@@ -1212,6 +1254,48 @@ paths:
             $ref: '#/definitions/ModelPublicServiceTriggerOrganizationModelBody'
       tags:
         - ModelPublicService
+  /v1alpha/{organization_model_name}/{version}/triggerAsync:
+    post:
+      summary: Trigger model inference asynchronously
+      description: |-
+        Triggers a deployed model to infer the result of a set of task or
+        questions.
+      operationId: ModelPublicService_TriggerAsyncOrganizationModel
+      responses:
+        "200":
+          description: A successful response.
+          schema:
+            $ref: '#/definitions/v1alphaTriggerAsyncOrganizationModelResponse'
+        "401":
+          description: Returned when the client credentials are not valid.
+          schema: {}
+        default:
+          description: An unexpected error response.
+          schema:
+            $ref: '#/definitions/googlerpcStatus'
+      parameters:
+        - name: organization_model_name
+          description: |-
+            The resource name of the model , which allows its access by parent organization
+            and ID.
+            - Format: `organizations/{organization.id}/models/{model.id}`.
+          in: path
+          required: true
+          type: string
+          pattern: organizations/[^/]+/models/[^/]+
+        - name: version
+          description: Model version tag
+          in: path
+          required: true
+          type: string
+          pattern: '[^/]+'
+        - name: body
+          in: body
+          required: true
+          schema:
+            $ref: '#/definitions/ModelPublicServiceTriggerAsyncOrganizationModelBody'
+      tags:
+        - ModelPublicService
   /v1alpha/{name}:
     get:
       summary: Get the details of a long-running operation
@@ -1283,6 +1367,34 @@ definitions:
     title: RenameUserModelRequest represents a request to rename a model
     required:
       - new_model_id
+  ModelPublicServiceTriggerAsyncOrganizationModelBody:
+    type: object
+    properties:
+      task_inputs:
+        type: array
+        items:
+          type: object
+          $ref: '#/definitions/v1alphaTaskInput'
+        description: Inference input parameters.
+    title: |-
+      TriggerAsyncOrganizationModelRequest represents a request to trigger a model inference
+      asynchronously
+    required:
+      - task_inputs
+  ModelPublicServiceTriggerAsyncUserModelBody:
+    type: object
+    properties:
+      task_inputs:
+        type: array
+        items:
+          type: object
+          $ref: '#/definitions/v1alphaTaskInput'
+        description: Inference input parameters.
+    description: |-
+      TriggerAsyncUserModelRequest represents a request to trigger a model inference
+      asynchronously.
+    required:
+      - task_inputs
   ModelPublicServiceTriggerOrganizationModelBody:
     type: object
     properties:
@@ -2740,6 +2852,26 @@ definitions:
         description: A list of generated images, encoded in base64.
         readOnly: true
     description: TextToImageOutput contains the result of a text-to-image task.
+  v1alphaTriggerAsyncOrganizationModelResponse:
+    type: object
+    properties:
+      operation:
+        $ref: '#/definitions/googlelongrunningOperation'
+        description: Long-running operation information.
+        readOnly: true
+    description: |-
+      TriggerAsyncOrganizationModelResponse contains the information to access the
+      status of an asynchronous model inference.
+  v1alphaTriggerAsyncUserModelResponse:
+    type: object
+    properties:
+      operation:
+        $ref: '#/definitions/googlelongrunningOperation'
+        description: Long-running operation information.
+        readOnly: true
+    description: |-
+      TriggerAsyncUserModelResponse contains the information to access the
+      status of an asynchronous model inference.
   v1alphaTriggerOrganizationModelBinaryFileUploadResponse:
     type: object
     properties: