[Vertex AI] Add a responseMimeType to generationConfig (#8235)

Provide Apps with an ability to configure the responseMimeType. This enables applications to request that VertexAI returns the response in a JSON format instead of a normal textual format.

Author: DellaBitta

common/api-review/vertexai-preview.api.md

@@ -311,6 +311,7 @@ export interface GenerationConfig {
     maxOutputTokens?: number;
     // (undocumented)
     presencePenalty?: number;
+    responseMimeType?: string;
     // (undocumented)
     stopSequences?: string[];
     // (undocumented)

docs-devsite/vertexai-preview.generationconfig.md

@@ -26,6 +26,7 @@ export interface GenerationConfig
 |  [frequencyPenalty](./vertexai-preview.generationconfig.md#generationconfigfrequencypenalty) | number |  |
 |  [maxOutputTokens](./vertexai-preview.generationconfig.md#generationconfigmaxoutputtokens) | number |  |
 |  [presencePenalty](./vertexai-preview.generationconfig.md#generationconfigpresencepenalty) | number |  |
+|  [responseMimeType](./vertexai-preview.generationconfig.md#generationconfigresponsemimetype) | string | Output response mimetype of the generated candidate text. Supported mimetype: <code>text/plain</code>: (default) Text output. <code>application/json</code>: JSON response in the candidates. The model needs to be prompted to output the appropriate response type, otherwise the behavior is undefined. This is a preview feature. |
 |  [stopSequences](./vertexai-preview.generationconfig.md#generationconfigstopsequences) | string\[\] |  |
 |  [temperature](./vertexai-preview.generationconfig.md#generationconfigtemperature) | number |  |
 |  [topK](./vertexai-preview.generationconfig.md#generationconfigtopk) | number |  |
@@ -63,6 +64,16 @@ maxOutputTokens?: number;
 presencePenalty?: number;
 ```
 
+## GenerationConfig.responseMimeType
+
+Output response mimetype of the generated candidate text. Supported mimetype: `text/plain`<!-- -->: (default) Text output. `application/json`<!-- -->: JSON response in the candidates. The model needs to be prompted to output the appropriate response type, otherwise the behavior is undefined. This is a preview feature.
+
+<b>Signature:</b>
+
+```typescript
+responseMimeType?: string;
+```
+
 ## GenerationConfig.stopSequences
 
 <b>Signature:</b>

packages/vertexai/src/types/requests.ts

@@ -77,6 +77,16 @@ export interface GenerationConfig {
   topK?: number;
   presencePenalty?: number;
   frequencyPenalty?: number;
+  /**
+   * Output response mimetype of the generated candidate text.
+   * Supported mimetype:
+   *   `text/plain`: (default) Text output.
+   *   `application/json`: JSON response in the candidates.
+   * The model needs to be prompted to output the appropriate response type,
+   * otherwise the behavior is undefined.
+   * This is a preview feature.
+   */
+  responseMimeType?: string;
 }
 
 /**