workers-ai-provider 0.4.0 → 0.5.0

package/README.md

@@ -19,10 +19,12 @@ binding = "AI"
 # ...
 ```
 
+### Using Workers AI
+
 Then in your Worker, import the factory function and create a new AI provider:
 
 ```ts
-import { createWorkersAI } from "../../../packages/workers-ai-provider/src";
+import { createWorkersAI } from "workers-ai-provider";
 import { streamText } from "ai";
 
 type Env = {
@@ -76,6 +78,45 @@ const text = await streamText({
 });
 ```
 
+### Using AutoRAG
+
+The provider now supports [Cloudflare's AutoRAG](https://developers.cloudflare.com/autorag/), allowing you to prompt your AutoRAG models directly from the Vercel AI SDK. Here's how to use it in your Worker:
+
+```ts
+import { createAutoRAG } from "workers-ai-provider";
+import { streamText } from "ai";
+
+type Env = {
+  AI: Ai;
+};
+
+export default {
+  async fetch(req: Request, env: Env) {
+    const autorag = createAutoRAG({ binding: env.AI.autorag('my-rag-name') });
+
+    const text = await streamText({
+      model: autorag("@cf/meta/llama-3.3-70b-instruct-fp8-fast"),
+      messages: [
+        {
+          role: "user",
+          content: "How to setup AI Gateway?",
+        },
+      ],
+    });
+
+    return text.toTextStreamResponse({
+      headers: {
+        // add these headers to ensure that the
+        // response is chunked and streamed
+        "Content-Type": "text/x-unknown",
+        "content-encoding": "identity",
+        "transfer-encoding": "chunked",
+      },
+    });
+  },
+};
+```
+
 For more info, refer to the documentation of the [Vercel AI SDK](https://sdk.vercel.ai/).
 
 ### Credits

package/dist/index.d.ts

@@ -1,20 +1,15 @@
-import { LanguageModelV1, ImageModelV1 } from '@ai-sdk/provider';
+import { LanguageModelV1, EmbeddingModelV1, ImageModelV1 } from '@ai-sdk/provider';
 
 type StringLike = string | {
     toString(): string;
 };
 
-type WorkersAIChatSettings = {
+type AutoRAGChatSettings = {
     /**
      * Whether to inject a safety prompt before all conversations.
      * Defaults to `false`.
      */
     safePrompt?: boolean;
-    /**
-     * Optionally set Cloudflare AI Gateway options.
-     * @deprecated
-     */
-    gateway?: GatewayOptions;
 } & {
     /**
      * Passthrough settings that are provided directly to the run function.
@@ -27,10 +22,84 @@ type WorkersAIChatSettings = {
  */
 type TextGenerationModels = Exclude<value2key<AiModels, BaseAiTextGeneration>, value2key<AiModels, BaseAiTextToImage>>;
 type ImageGenerationModels = value2key<AiModels, BaseAiTextToImage>;
+/**
+ * The names of the BaseAiTextToEmbeddings models.
+ */
+type EmbeddingModels = value2key<AiModels, BaseAiTextEmbeddings>;
 type value2key<T, V> = {
     [K in keyof T]: T[K] extends V ? K : never;
 }[keyof T];
 
+type AutoRAGChatConfig = {
+    provider: string;
+    binding: AutoRAG;
+    gateway?: GatewayOptions;
+};
+declare class AutoRAGChatLanguageModel implements LanguageModelV1 {
+    readonly specificationVersion = "v1";
+    readonly defaultObjectGenerationMode = "json";
+    readonly modelId: TextGenerationModels;
+    readonly settings: AutoRAGChatSettings;
+    private readonly config;
+    constructor(modelId: TextGenerationModels, settings: AutoRAGChatSettings, config: AutoRAGChatConfig);
+    get provider(): string;
+    private getArgs;
+    doGenerate(options: Parameters<LanguageModelV1["doGenerate"]>[0]): Promise<Awaited<ReturnType<LanguageModelV1["doGenerate"]>>>;
+    doStream(options: Parameters<LanguageModelV1["doStream"]>[0]): Promise<Awaited<ReturnType<LanguageModelV1["doStream"]>>>;
+}
+
+type WorkersAIEmbeddingConfig = {
+    provider: string;
+    binding: Ai;
+    gateway?: GatewayOptions;
+};
+type WorkersAIEmbeddingSettings = {
+    gateway?: GatewayOptions;
+    maxEmbeddingsPerCall?: number;
+    supportsParallelCalls?: boolean;
+} & {
+    /**
+     * Arbitrary provider-specific options forwarded unmodified.
+     */
+    [key: string]: StringLike;
+};
+declare class WorkersAIEmbeddingModel implements EmbeddingModelV1<string> {
+    /**
+     * Semantic version of the {@link EmbeddingModelV1} specification implemented
+     * by this class. It never changes.
+     */
+    readonly specificationVersion = "v1";
+    readonly modelId: EmbeddingModels;
+    private readonly config;
+    private readonly settings;
+    /**
+     * Provider name exposed for diagnostics and error reporting.
+     */
+    get provider(): string;
+    get maxEmbeddingsPerCall(): number;
+    get supportsParallelCalls(): boolean;
+    constructor(modelId: EmbeddingModels, settings: WorkersAIEmbeddingSettings, config: WorkersAIEmbeddingConfig);
+    doEmbed({ values, }: Parameters<EmbeddingModelV1<string>["doEmbed"]>[0]): Promise<Awaited<ReturnType<EmbeddingModelV1<string>["doEmbed"]>>>;
+}
+
+type WorkersAIChatSettings = {
+    /**
+     * Whether to inject a safety prompt before all conversations.
+     * Defaults to `false`.
+     */
+    safePrompt?: boolean;
+    /**
+     * Optionally set Cloudflare AI Gateway options.
+     * @deprecated
+     */
+    gateway?: GatewayOptions;
+} & {
+    /**
+     * Passthrough settings that are provided directly to the run function.
+     */
+    [key: string]: StringLike;
+};
+
 type WorkersAIChatConfig = {
     provider: string;
     binding: Ai;
@@ -102,6 +171,9 @@ interface WorkersAI {
      * Creates a model for text generation.
      **/
     chat(modelId: TextGenerationModels, settings?: WorkersAIChatSettings): WorkersAIChatLanguageModel;
+    embedding(modelId: EmbeddingModels, settings?: WorkersAIEmbeddingSettings): WorkersAIEmbeddingModel;
+    textEmbedding(modelId: EmbeddingModels, settings?: WorkersAIEmbeddingSettings): WorkersAIEmbeddingModel;
+    textEmbeddingModel(modelId: EmbeddingModels, settings?: WorkersAIEmbeddingSettings): WorkersAIEmbeddingModel;
     /**
      * Creates a model for image generation.
      **/
@@ -111,5 +183,19 @@ interface WorkersAI {
  * Create a Workers AI provider instance.
  */
 declare function createWorkersAI(options: WorkersAISettings): WorkersAI;
+type AutoRAGSettings = {
+    binding: AutoRAG;
+};
+interface AutoRAGProvider {
+    (options?: AutoRAGChatSettings): AutoRAGChatLanguageModel;
+    /**
+     * Creates a model for text generation.
+     **/
+    chat(settings?: AutoRAGChatSettings): AutoRAGChatLanguageModel;
+}
+/**
+ * Create a Workers AI provider instance.
+ */
+declare function createAutoRAG(options: AutoRAGSettings): AutoRAGProvider;
 
-export { type WorkersAI, type WorkersAISettings, createWorkersAI };
+export { type AutoRAGProvider, type AutoRAGSettings, type WorkersAI, type WorkersAISettings, createAutoRAG, createWorkersAI };