@friendliai/ai-provider 0.1.1

package/README.md

@@ -0,0 +1,44 @@
+# @friendliai/ai-provider
+
+Vercel AI SDK Provider for FriendliAI
+
+## Installation
+
+You can install the package via npm:
+
+```bash
+npm i @friendliai/ai-provider
+```
+
+## Provider Instance
+
+You can import the default provider instance `friendliai` from `@friendliai/ai-provider`:
+
+```ts
+import { friendliai } from "@friendliai/ai-provider";
+```
+
+If you need a customized setup, you can import `createFriendliAI` from `@friendliai/ai-provider` and create a provider instance with your settings:
+
+```ts
+import { createFriendliAI } from "@friendliai/ai-provider";
+
+const friendliai = createFriendliAI({
+  // custom settings
+});
+```
+
+You can use the following optional settings to customize the FriendliAI provider instance:
+
+- **baseURL** _string_
+
+  Use a different URL prefix for API calls, e.g. to use proxy servers.
+  The default prefix is `https://inference.friendli.ai/v1`.
+
+- **teamId** _string_
+
+  The team ID to use for the requests.
+
+- **headers** _Record<string,string>_
+
+  Custom headers to include in the requests.

package/dist/index.d.mts

@@ -0,0 +1,257 @@
+import { LanguageModelV1, ProviderV1 } from '@ai-sdk/provider';
+import { FetchFunction } from '@ai-sdk/provider-utils';
+
+type FriendliAIChatModelId = "meta-llama-3.1-8b-instruct" | "meta-llama-3.1-70b-instruct" | "mixtral-8x7b-instruct-v0-1" | (string & {});
+interface FriendliAIChatSettings {
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+  */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to better understand how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+  */
+    logprobs?: boolean | number;
+    /**
+  Whether to enable parallel function calling during tool use. Default to true.
+     */
+    parallelToolCalls?: boolean;
+    /**
+  Whether to use structured outputs. Defaults to false.
+  
+  When enabled, tool calls and object generation will be strict and follow the provided schema.
+   */
+    structuredOutputs?: boolean;
+    /**
+     BETA FEATURE: Include the model's training loss in the response.
+     */
+    tools?: Array<{
+        type: "web:site" | "web:search" | "math:calendar" | "math:statistics" | "math:calculator" | "code:python-interpreter";
+    }>;
+    /**
+  Whether to use legacy function calling. Defaults to false.
+  
+  Required by some open source inference engines which do not support the `tools` API. May also
+  provide a workaround for `parallelToolCalls` resulting in the provider buffering tool calls,
+  which causes `streamObject` to be non-streaming.
+  
+  Prefer setting `parallelToolCalls: false` over this option.
+  
+  @deprecated this API is supported but deprecated by FriendliAI.
+     */
+    useLegacyFunctionCalling?: boolean;
+    /**
+  A unique identifier representing your end-user, which can help FriendliAI to
+  monitor and detect abuse. Learn more.
+  */
+    user?: string;
+}
+
+type FriendliAIChatConfig = {
+    provider: string;
+    compatibility: "strict" | "compatible";
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: FetchFunction;
+};
+declare class FriendliAIChatLanguageModel implements LanguageModelV1 {
+    readonly specificationVersion = "v1";
+    readonly modelId: FriendliAIChatModelId;
+    readonly settings: FriendliAIChatSettings;
+    private readonly config;
+    constructor(modelId: FriendliAIChatModelId, settings: FriendliAIChatSettings, config: FriendliAIChatConfig);
+    get supportsStructuredOutputs(): boolean;
+    get defaultObjectGenerationMode(): "tool" | "json";
+    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 FriendliAICompletionModelId = "meta-llama-3.1-70b-instruct" | "meta-llama-3.1-8b-instruct" | "mixtral-8x7b-instruct-v0-1" | (string & {});
+interface FriendliAICompletionSettings {
+    /**
+  Echo back the prompt in addition to the completion.
+     */
+    echo?: boolean;
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+     */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to better understand how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+     */
+    logprobs?: boolean | number;
+    /**
+  The suffix that comes after a completion of inserted text.
+     */
+    suffix?: string;
+    /**
+  A unique identifier representing your end-user, which can help FriendliAI to
+  monitor and detect abuse. Learn more.
+     */
+    user?: string;
+}
+
+type FriendliAICompletionConfig = {
+    provider: string;
+    compatibility: "strict" | "compatible";
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: FetchFunction;
+};
+declare class FriendliAICompletionLanguageModel implements LanguageModelV1 {
+    readonly specificationVersion = "v1";
+    readonly defaultObjectGenerationMode: undefined;
+    readonly modelId: FriendliAICompletionModelId;
+    readonly settings: FriendliAICompletionSettings;
+    private readonly config;
+    constructor(modelId: FriendliAICompletionModelId, settings: FriendliAICompletionSettings, config: FriendliAICompletionConfig);
+    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"]>>>;
+}
+
+interface FriendliAIProvider extends ProviderV1 {
+    (modelId: "meta-llama-3.1-8b-instruct", settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+    (modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI model for text generation.
+     */
+    languageModel(modelId: "meta-llama-3.1-8b-instruct", settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+    languageModel(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI chat model for text generation.
+     */
+    chat(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI completion model for text generation.
+     */
+    completion(modelId: FriendliAICompletionModelId, settings?: FriendliAICompletionSettings): LanguageModelV1;
+}
+interface FriendliAIProviderSettings {
+    /**
+  Base URL for the FriendliAI API calls.
+       */
+    baseURL?: string;
+    /**
+  @deprecated Use `baseURL` instead.
+       */
+    baseUrl?: string;
+    /**
+  API key for authenticating requests.
+       */
+    apiKey?: string;
+    /**
+     FriendliAI Team ID.
+     */
+    teamId?: string;
+    /**
+  FriendliAI project.
+       */
+    project?: string;
+    /**
+  Custom headers to include in the requests.
+       */
+    headers?: Record<string, string>;
+    /**
+  FriendliAI compatibility mode. Should be set to `strict` when using the FriendliAI API,
+  and `compatible` when using 3rd party providers. In `compatible` mode, newer
+  information such as streamOptions are not being sent. Defaults to 'compatible'.
+     */
+    compatibility?: "strict" | "compatible";
+    /**
+  Custom fetch implementation. You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+      */
+    fetch?: FetchFunction;
+}
+/**
+Create an FriendliAI provider instance.
+ */
+declare function createFriendliAI(options?: FriendliAIProviderSettings): FriendliAIProvider;
+/**
+Default FriendliAI provider instance. It uses 'strict' compatibility mode.
+ */
+declare const friendliai: FriendliAIProvider;
+
+/**
+@deprecated Use `createFriendliAI` instead.
+ */
+declare class FriendliAI {
+    /**
+  Use a different URL prefix for API calls, e.g. to use proxy servers.
+  The default prefix is `https://api.friendliai.com/v1`.
+     */
+    readonly baseURL: string;
+    /**
+  API key that is being send using the `Authorization` header.
+  It defaults to the `FRIEDNLI_API_KEY` environment variable.
+   */
+    readonly apiKey?: string;
+    /**
+     FriendliAI Team ID.
+     */
+    teamId?: string;
+    /**
+  FriendliAI project.
+     */
+    readonly project?: string;
+    /**
+  Custom headers to include in the requests.
+     */
+    readonly headers?: Record<string, string>;
+    /**
+     * Creates a new FriendliAI provider instance.
+     */
+    constructor(options?: FriendliAIProviderSettings);
+    private get baseConfig();
+    chat(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): FriendliAIChatLanguageModel;
+    completion(modelId: FriendliAICompletionModelId, settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+}
+
+export { FriendliAI, type FriendliAIProvider, type FriendliAIProviderSettings, createFriendliAI, friendliai };

package/dist/index.d.ts

@@ -0,0 +1,257 @@
+import { LanguageModelV1, ProviderV1 } from '@ai-sdk/provider';
+import { FetchFunction } from '@ai-sdk/provider-utils';
+
+type FriendliAIChatModelId = "meta-llama-3.1-8b-instruct" | "meta-llama-3.1-70b-instruct" | "mixtral-8x7b-instruct-v0-1" | (string & {});
+interface FriendliAIChatSettings {
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+  */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to better understand how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+  */
+    logprobs?: boolean | number;
+    /**
+  Whether to enable parallel function calling during tool use. Default to true.
+     */
+    parallelToolCalls?: boolean;
+    /**
+  Whether to use structured outputs. Defaults to false.
+  
+  When enabled, tool calls and object generation will be strict and follow the provided schema.
+   */
+    structuredOutputs?: boolean;
+    /**
+     BETA FEATURE: Include the model's training loss in the response.
+     */
+    tools?: Array<{
+        type: "web:site" | "web:search" | "math:calendar" | "math:statistics" | "math:calculator" | "code:python-interpreter";
+    }>;
+    /**
+  Whether to use legacy function calling. Defaults to false.
+  
+  Required by some open source inference engines which do not support the `tools` API. May also
+  provide a workaround for `parallelToolCalls` resulting in the provider buffering tool calls,
+  which causes `streamObject` to be non-streaming.
+  
+  Prefer setting `parallelToolCalls: false` over this option.
+  
+  @deprecated this API is supported but deprecated by FriendliAI.
+     */
+    useLegacyFunctionCalling?: boolean;
+    /**
+  A unique identifier representing your end-user, which can help FriendliAI to
+  monitor and detect abuse. Learn more.
+  */
+    user?: string;
+}
+
+type FriendliAIChatConfig = {
+    provider: string;
+    compatibility: "strict" | "compatible";
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: FetchFunction;
+};
+declare class FriendliAIChatLanguageModel implements LanguageModelV1 {
+    readonly specificationVersion = "v1";
+    readonly modelId: FriendliAIChatModelId;
+    readonly settings: FriendliAIChatSettings;
+    private readonly config;
+    constructor(modelId: FriendliAIChatModelId, settings: FriendliAIChatSettings, config: FriendliAIChatConfig);
+    get supportsStructuredOutputs(): boolean;
+    get defaultObjectGenerationMode(): "tool" | "json";
+    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 FriendliAICompletionModelId = "meta-llama-3.1-70b-instruct" | "meta-llama-3.1-8b-instruct" | "mixtral-8x7b-instruct-v0-1" | (string & {});
+interface FriendliAICompletionSettings {
+    /**
+  Echo back the prompt in addition to the completion.
+     */
+    echo?: boolean;
+    /**
+  Modify the likelihood of specified tokens appearing in the completion.
+  
+  Accepts a JSON object that maps tokens (specified by their token ID in
+  the GPT tokenizer) to an associated bias value from -100 to 100. You
+  can use this tokenizer tool to convert text to token IDs. Mathematically,
+  the bias is added to the logits generated by the model prior to sampling.
+  The exact effect will vary per model, but values between -1 and 1 should
+  decrease or increase likelihood of selection; values like -100 or 100
+  should result in a ban or exclusive selection of the relevant token.
+  
+  As an example, you can pass {"50256": -100} to prevent the <|endoftext|>
+  token from being generated.
+     */
+    logitBias?: Record<number, number>;
+    /**
+  Return the log probabilities of the tokens. Including logprobs will increase
+  the response size and can slow down response times. However, it can
+  be useful to better understand how the model is behaving.
+  
+  Setting to true will return the log probabilities of the tokens that
+  were generated.
+  
+  Setting to a number will return the log probabilities of the top n
+  tokens that were generated.
+     */
+    logprobs?: boolean | number;
+    /**
+  The suffix that comes after a completion of inserted text.
+     */
+    suffix?: string;
+    /**
+  A unique identifier representing your end-user, which can help FriendliAI to
+  monitor and detect abuse. Learn more.
+     */
+    user?: string;
+}
+
+type FriendliAICompletionConfig = {
+    provider: string;
+    compatibility: "strict" | "compatible";
+    headers: () => Record<string, string | undefined>;
+    url: (options: {
+        modelId: string;
+        path: string;
+    }) => string;
+    fetch?: FetchFunction;
+};
+declare class FriendliAICompletionLanguageModel implements LanguageModelV1 {
+    readonly specificationVersion = "v1";
+    readonly defaultObjectGenerationMode: undefined;
+    readonly modelId: FriendliAICompletionModelId;
+    readonly settings: FriendliAICompletionSettings;
+    private readonly config;
+    constructor(modelId: FriendliAICompletionModelId, settings: FriendliAICompletionSettings, config: FriendliAICompletionConfig);
+    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"]>>>;
+}
+
+interface FriendliAIProvider extends ProviderV1 {
+    (modelId: "meta-llama-3.1-8b-instruct", settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+    (modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI model for text generation.
+     */
+    languageModel(modelId: "meta-llama-3.1-8b-instruct", settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+    languageModel(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI chat model for text generation.
+     */
+    chat(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): LanguageModelV1;
+    /**
+  Creates an FriendliAI completion model for text generation.
+     */
+    completion(modelId: FriendliAICompletionModelId, settings?: FriendliAICompletionSettings): LanguageModelV1;
+}
+interface FriendliAIProviderSettings {
+    /**
+  Base URL for the FriendliAI API calls.
+       */
+    baseURL?: string;
+    /**
+  @deprecated Use `baseURL` instead.
+       */
+    baseUrl?: string;
+    /**
+  API key for authenticating requests.
+       */
+    apiKey?: string;
+    /**
+     FriendliAI Team ID.
+     */
+    teamId?: string;
+    /**
+  FriendliAI project.
+       */
+    project?: string;
+    /**
+  Custom headers to include in the requests.
+       */
+    headers?: Record<string, string>;
+    /**
+  FriendliAI compatibility mode. Should be set to `strict` when using the FriendliAI API,
+  and `compatible` when using 3rd party providers. In `compatible` mode, newer
+  information such as streamOptions are not being sent. Defaults to 'compatible'.
+     */
+    compatibility?: "strict" | "compatible";
+    /**
+  Custom fetch implementation. You can use it as a middleware to intercept requests,
+  or to provide a custom fetch implementation for e.g. testing.
+      */
+    fetch?: FetchFunction;
+}
+/**
+Create an FriendliAI provider instance.
+ */
+declare function createFriendliAI(options?: FriendliAIProviderSettings): FriendliAIProvider;
+/**
+Default FriendliAI provider instance. It uses 'strict' compatibility mode.
+ */
+declare const friendliai: FriendliAIProvider;
+
+/**
+@deprecated Use `createFriendliAI` instead.
+ */
+declare class FriendliAI {
+    /**
+  Use a different URL prefix for API calls, e.g. to use proxy servers.
+  The default prefix is `https://api.friendliai.com/v1`.
+     */
+    readonly baseURL: string;
+    /**
+  API key that is being send using the `Authorization` header.
+  It defaults to the `FRIEDNLI_API_KEY` environment variable.
+   */
+    readonly apiKey?: string;
+    /**
+     FriendliAI Team ID.
+     */
+    teamId?: string;
+    /**
+  FriendliAI project.
+     */
+    readonly project?: string;
+    /**
+  Custom headers to include in the requests.
+     */
+    readonly headers?: Record<string, string>;
+    /**
+     * Creates a new FriendliAI provider instance.
+     */
+    constructor(options?: FriendliAIProviderSettings);
+    private get baseConfig();
+    chat(modelId: FriendliAIChatModelId, settings?: FriendliAIChatSettings): FriendliAIChatLanguageModel;
+    completion(modelId: FriendliAICompletionModelId, settings?: FriendliAICompletionSettings): FriendliAICompletionLanguageModel;
+}
+
+export { FriendliAI, type FriendliAIProvider, type FriendliAIProviderSettings, createFriendliAI, friendliai };