This is comparable to [GenerativeModel.generateContent()](./ai.generativemodel.md#generativemodelgeneratecontent) for generating content using in-cloud inference.
| -| [generateContentStream(request)](./ai.chromeadapter.md#chromeadaptergeneratecontentstream) | Generates a content stream using on-device inference.This is comparable to [GenerativeModel.generateContentStream()](./ai.generativemodel.md#generativemodelgeneratecontentstream) for generating a content stream using in-cloud inference.
| +| [generateContent(request)](./ai.chromeadapter.md#chromeadaptergeneratecontent) | Generates content using on-device inference. | +| [generateContentStream(request)](./ai.chromeadapter.md#chromeadaptergeneratecontentstream) | Generates a content stream using on-device inference. | | [isAvailable(request)](./ai.chromeadapter.md#chromeadapterisavailable) | Checks if the on-device model is capable of handling a given request. | ## ChromeAdapter.generateContent() Generates content using on-device inference. -This is comparable to [GenerativeModel.generateContent()](./ai.generativemodel.md#generativemodelgeneratecontent) for generating content using in-cloud inference.
+This is comparable to [GenerativeModel.generateContent()](./ai.generativemodel.md#generativemodelgeneratecontent) for generating content using in-cloud inference. Signature: @@ -54,7 +54,7 @@ Promise<Response> Generates a content stream using on-device inference. -This is comparable to [GenerativeModel.generateContentStream()](./ai.generativemodel.md#generativemodelgeneratecontentstream) for generating a content stream using in-cloud inference.
+This is comparable to [GenerativeModel.generateContentStream()](./ai.generativemodel.md#generativemodelgeneratecontentstream) for generating a content stream using in-cloud inference. Signature: diff --git a/docs-devsite/ai.requestoptions.md b/docs-devsite/ai.requestoptions.md index 8178ef5b696..c04230fcd62 100644 --- a/docs-devsite/ai.requestoptions.md +++ b/docs-devsite/ai.requestoptions.md @@ -22,12 +22,12 @@ export interface RequestOptions | Property | Type | Description | | --- | --- | --- | -| [baseUrl](./ai.requestoptions.md#requestoptionsbaseurl) | string | Base url for endpoint. Defaults to https://firebasevertexai.googleapis.com | +| [baseUrl](./ai.requestoptions.md#requestoptionsbaseurl) | string | Base url for endpoint. Defaults to https://firebasevertexai.googleapis.com, which is the [Firebase AI Logic API](https://console.cloud.google.com/apis/library/firebasevertexai.googleapis.com?project=_) (used regardless of your chosen Gemini API provider). | | [timeout](./ai.requestoptions.md#requestoptionstimeout) | number | Request timeout in milliseconds. Defaults to 180 seconds (180000ms). | ## RequestOptions.baseUrl -Base url for endpoint. Defaults to https://firebasevertexai.googleapis.com +Base url for endpoint. Defaults to https://firebasevertexai.googleapis.com, which is the [Firebase AI Logic API](https://console.cloud.google.com/apis/library/firebasevertexai.googleapis.com?project=_) (used regardless of your chosen Gemini API provider). Signature: diff --git a/packages/ai/src/methods/chrome-adapter.ts b/packages/ai/src/methods/chrome-adapter.ts index 4dea4170c0d..8fc4b23e665 100644 --- a/packages/ai/src/methods/chrome-adapter.ts +++ b/packages/ai/src/methods/chrome-adapter.ts @@ -61,17 +61,17 @@ export class ChromeAdapterImpl implements ChromeAdapter { /** * Checks if a given request can be made on-device. * - *Pros: callers needn't be concerned with details of on-device availability.
- *Cons: this method spans a few concerns and splits request validation from usage. + * + * Pros: callers needn't be concerned with details of on-device availability.
+ * Cons: this method spans a few concerns and splits request validation from usage. * If instance variables weren't already part of the API, we could consider a better - * separation of concerns. + * separation of concerns. */ async isAvailable(request: GenerateContentRequest): PromiseThis is comparable to {@link GenerativeModel.generateContent} for generating content in - * Cloud.
+ * @remarks + * This is comparable to {@link GenerativeModel.generateContent} for generating content in + * Cloud. * @param request - a standard Firebase AI {@link GenerateContentRequest} * @returns {@link Response}, so we can reuse common response formatting. */ @@ -149,8 +150,9 @@ export class ChromeAdapterImpl implements ChromeAdapter { /** * Generates content stream on device. * - *This is comparable to {@link GenerativeModel.generateContentStream} for generating content in - * Cloud.
+ * @remarks + * This is comparable to {@link GenerativeModel.generateContentStream} for generating content in + * Cloud. * @param request - a standard Firebase AI {@link GenerateContentRequest} * @returns {@link Response}, so we can reuse common response formatting. */ @@ -228,11 +230,11 @@ export class ChromeAdapterImpl implements ChromeAdapter { /** * Triggers out-of-band download of an on-device model. * - *Chrome only downloads models as needed. Chrome knows a model is needed when code calls - * LanguageModel.create.
+ * Chrome only downloads models as needed. Chrome knows a model is needed when code calls + * LanguageModel.create. * - *Since Chrome manages the download, the SDK can only avoid redundant download requests by - * tracking if a download has previously been requested.
+ * Since Chrome manages the download, the SDK can only avoid redundant download requests by + * tracking if a download has previously been requested. */ private download(): void { if (this.isDownloading) { @@ -302,12 +304,12 @@ export class ChromeAdapterImpl implements ChromeAdapter { /** * Abstracts Chrome session creation. * - *Chrome uses a multi-turn session for all inference. Firebase AI uses single-turn for all + * Chrome uses a multi-turn session for all inference. Firebase AI uses single-turn for all * inference. To map the Firebase AI API to Chrome's API, the SDK creates a new session for all - * inference.
+ * inference. * - *Chrome will remove a model from memory if it's no longer in use, so this method ensures a - * new session is created before an old session is destroyed.
+ * Chrome will remove a model from memory if it's no longer in use, so this method ensures a + * new session is created before an old session is destroyed. */ private async createSession(): PromiseThis is comparable to {@link GenerativeModel.generateContent} for generating - * content using in-cloud inference.
+ * @remarks + * This is comparable to {@link GenerativeModel.generateContent} for generating + * content using in-cloud inference. * @param request - a standard Firebase AI {@link GenerateContentRequest} */ generateContent(request: GenerateContentRequest): PromiseThis is comparable to {@link GenerativeModel.generateContentStream} for generating - * a content stream using in-cloud inference.
+ * @remarks + * This is comparable to {@link GenerativeModel.generateContentStream} for generating + * a content stream using in-cloud inference. * @param request - a standard Firebase AI {@link GenerateContentRequest} */ generateContentStream(request: GenerateContentRequest): Promise