509 lines
14 KiB
JavaScript
509 lines
14 KiB
JavaScript
import { ChatOpenAI, } from "@langchain/openai";
|
|
import { getEnvironmentVariable } from "@langchain/core/utils/env";
|
|
/**
|
|
* Wrapper around Fireworks API for large language models fine-tuned for chat
|
|
*
|
|
* Fireworks API is compatible to the OpenAI API with some limitations described in
|
|
* https://readme.fireworks.ai/docs/openai-compatibility.
|
|
*
|
|
* To use, you should have the `FIREWORKS_API_KEY` environment variable set.
|
|
*
|
|
* Setup:
|
|
* Install `@langchain/community` and set a environment variable called `FIREWORKS_API_KEY`.
|
|
*
|
|
* ```bash
|
|
* npm install @langchain/community
|
|
* export FIREWORKS_API_KEY="your-api-key"
|
|
* ```
|
|
*
|
|
* ## [Constructor args](https://api.js.langchain.com/classes/langchain_community_chat_models_fireworks.ChatFireworks.html#constructor)
|
|
*
|
|
* ## [Runtime args](https://api.js.langchain.com/interfaces/langchain_openai.ChatOpenAICallOptions.html)
|
|
*
|
|
* Because the Fireworks API extends OpenAI's, the call option type is the same.
|
|
*
|
|
* Runtime args can be passed as the second argument to any of the base runnable methods `.invoke`. `.stream`, `.batch`, etc.
|
|
* They can also be passed via `.bind`, or the second arg in `.bindTools`, like shown in the examples below:
|
|
*
|
|
* ```typescript
|
|
* // When calling `.bind`, call options should be passed via the first argument
|
|
* const llmWithArgsBound = llm.bind({
|
|
* stop: ["\n"],
|
|
* tools: [...],
|
|
* });
|
|
*
|
|
* // When calling `.bindTools`, call options should be passed via the second argument
|
|
* const llmWithTools = llm.bindTools(
|
|
* [...],
|
|
* {
|
|
* stop: ["\n"],
|
|
* }
|
|
* );
|
|
* ```
|
|
*
|
|
* ## Examples
|
|
*
|
|
* <details open>
|
|
* <summary><strong>Instantiate</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* import { ChatFireworks } from '@langchain/community/chat_models/fireworks';
|
|
*
|
|
* const llm = new ChatFireworks({
|
|
* model: "command-r-plus",
|
|
* temperature: 0,
|
|
* // other params...
|
|
* });
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Invoking</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* const input = `Translate "I love programming" into French.`;
|
|
*
|
|
* // Models also accept a list of chat messages or a formatted prompt
|
|
* const result = await llm.invoke(input);
|
|
* console.log(result);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* AIMessage {
|
|
* "id": "dbc233df-532e-4aaa-8995-9d6ea65fea15",
|
|
* "content": "The translation of \"I love programming\" into French is:\n\n\"J'adore la programmation.\"\n\nHere's a breakdown of the translation:\n\n* \"I\" is translated to \"Je\" (but in informal writing, it's common to use \"J'\" instead of \"Je\" when it's followed by a vowel)\n* \"love\" is translated to \"adore\"\n* \"programming\" is translated to \"la programmation\"\n\nSo, the complete translation is \"J'adore la programmation.\"",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "tokenUsage": {
|
|
* "completionTokens": 105,
|
|
* "promptTokens": 19,
|
|
* "totalTokens": 124
|
|
* },
|
|
* "finish_reason": "stop"
|
|
* },
|
|
* "tool_calls": [],
|
|
* "invalid_tool_calls": [],
|
|
* "usage_metadata": {
|
|
* "input_tokens": 19,
|
|
* "output_tokens": 105,
|
|
* "total_tokens": 124
|
|
* }
|
|
* }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Streaming Chunks</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* for await (const chunk of await llm.stream(input)) {
|
|
* console.log(chunk);
|
|
* }
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "The translation",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": " of \"",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "I love",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": " programming\"",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": " into French",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": " is:\n\n",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "\"J",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* ...
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "ation.\"",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": null
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "id": "ed5fc403-b7ed-4447-819f-f9645ea0277c",
|
|
* "content": "",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": "stop"
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": []
|
|
* }
|
|
* AIMessageChunk {
|
|
* "content": "",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {},
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": [],
|
|
* "usage_metadata": {
|
|
* "input_tokens": 19,
|
|
* "output_tokens": 105,
|
|
* "total_tokens": 124
|
|
* }
|
|
* }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Aggregate Streamed Chunks</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* import { AIMessageChunk } from '@langchain/core/messages';
|
|
* import { concat } from '@langchain/core/utils/stream';
|
|
*
|
|
* const stream = await llm.stream(input);
|
|
* let full: AIMessageChunk | undefined;
|
|
* for await (const chunk of stream) {
|
|
* full = !full ? chunk : concat(full, chunk);
|
|
* }
|
|
* console.log(full);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* AIMessageChunk {
|
|
* "id": "9b80e5af-0f50-4fb7-b700-6d431a819556",
|
|
* "content": "The translation of \"I love programming\" into French is:\n\n\"J'adore la programmation.\"\n\nHere's a breakdown of the translation:\n\n* \"I\" is translated to \"Je\" (but in informal writing, it's common to use \"J'\" instead of \"Je\" when it's followed by a vowel)\n* \"love\" is translated to \"adore\"\n* \"programming\" is translated to \"la programmation\"\n\nSo, the complete translation is \"J'adore la programmation.\"",
|
|
* "additional_kwargs": {},
|
|
* "response_metadata": {
|
|
* "prompt": 0,
|
|
* "completion": 0,
|
|
* "finish_reason": "stop"
|
|
* },
|
|
* "tool_calls": [],
|
|
* "tool_call_chunks": [],
|
|
* "invalid_tool_calls": [],
|
|
* "usage_metadata": {
|
|
* "input_tokens": 19,
|
|
* "output_tokens": 105,
|
|
* "total_tokens": 124
|
|
* }
|
|
* }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Bind tools</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* import { z } from 'zod';
|
|
*
|
|
* const llmForToolCalling = new ChatFireworks({
|
|
* // Use a model with tool calling capability
|
|
* model: "accounts/fireworks/models/firefunction-v2",
|
|
* temperature: 0,
|
|
* // other params...
|
|
* });
|
|
* const GetWeather = {
|
|
* name: "GetWeather",
|
|
* description: "Get the current weather in a given location",
|
|
* schema: z.object({
|
|
* location: z.string().describe("The city and state, e.g. San Francisco, CA")
|
|
* }),
|
|
* }
|
|
*
|
|
* const GetPopulation = {
|
|
* name: "GetPopulation",
|
|
* description: "Get the current population in a given location",
|
|
* schema: z.object({
|
|
* location: z.string().describe("The city and state, e.g. San Francisco, CA")
|
|
* }),
|
|
* }
|
|
*
|
|
* const llmWithTools = llmForToolCalling.bindTools([GetWeather, GetPopulation]);
|
|
* const aiMsg = await llmWithTools.invoke(
|
|
* "Which city is hotter today and which is bigger: LA or NY?"
|
|
* );
|
|
* console.log(aiMsg.tool_calls);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* [
|
|
* {
|
|
* name: 'GetWeather',
|
|
* args: { location: 'Los Angeles, CA' },
|
|
* type: 'tool_call',
|
|
* id: 'call_9DE0WnhgKDbxu6HyHOkDQFub'
|
|
* },
|
|
* {
|
|
* name: 'GetWeather',
|
|
* args: { location: 'New York, NY' },
|
|
* type: 'tool_call',
|
|
* id: 'call_58lcAPTqQyiqepxynwARhGs8'
|
|
* },
|
|
* {
|
|
* name: 'GetPopulation',
|
|
* args: { location: 'Los Angeles, CA' },
|
|
* type: 'tool_call',
|
|
* id: 'call_r0m6AFoqaMvPp4Zt5aEAc0oE'
|
|
* },
|
|
* {
|
|
* name: 'GetPopulation',
|
|
* args: { location: 'New York, NY' },
|
|
* type: 'tool_call',
|
|
* id: 'call_mENaPG1ryOF44BmaW4VkBaSi'
|
|
* }
|
|
* ]
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Structured Output</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* import { z } from 'zod';
|
|
*
|
|
* const Joke = z.object({
|
|
* setup: z.string().describe("The setup of the joke"),
|
|
* punchline: z.string().describe("The punchline to the joke"),
|
|
* rating: z.number().optional().describe("How funny the joke is, from 1 to 10")
|
|
* }).describe('Joke to tell user.');
|
|
*
|
|
* const structuredLlm = llmForToolCalling.withStructuredOutput(Joke, { name: "Joke" });
|
|
* const jokeResult = await structuredLlm.invoke("Tell me a joke about cats");
|
|
* console.log(jokeResult);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* {
|
|
* setup: 'Why did the cat join a band?',
|
|
* punchline: 'Because it wanted to be the purr-cussionist!',
|
|
* rating: 8
|
|
* }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
*
|
|
* <summary><strong>Usage Metadata</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* const aiMsgForMetadata = await llm.invoke(input);
|
|
* console.log(aiMsgForMetadata.usage_metadata);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* { input_tokens: 277, output_tokens: 8, total_tokens: 285 }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*
|
|
* <details>
|
|
* <summary><strong>Response Metadata</strong></summary>
|
|
*
|
|
* ```typescript
|
|
* const aiMsgForResponseMetadata = await llm.invoke(input);
|
|
* console.log(aiMsgForResponseMetadata.response_metadata);
|
|
* ```
|
|
*
|
|
* ```txt
|
|
* {
|
|
* tokenUsage: { completionTokens: 8, promptTokens: 277, totalTokens: 285 },
|
|
* finish_reason: 'stop'
|
|
* }
|
|
* ```
|
|
* </details>
|
|
*
|
|
* <br />
|
|
*/
|
|
export class ChatFireworks extends ChatOpenAI {
|
|
static lc_name() {
|
|
return "ChatFireworks";
|
|
}
|
|
_llmType() {
|
|
return "fireworks";
|
|
}
|
|
get lc_secrets() {
|
|
return {
|
|
fireworksApiKey: "FIREWORKS_API_KEY",
|
|
apiKey: "FIREWORKS_API_KEY",
|
|
};
|
|
}
|
|
constructor(fields) {
|
|
const fireworksApiKey = fields?.apiKey ||
|
|
fields?.fireworksApiKey ||
|
|
getEnvironmentVariable("FIREWORKS_API_KEY");
|
|
if (!fireworksApiKey) {
|
|
throw new Error(`Fireworks API key not found. Please set the FIREWORKS_API_KEY environment variable or provide the key into "fireworksApiKey"`);
|
|
}
|
|
super({
|
|
...fields,
|
|
model: fields?.model ||
|
|
fields?.modelName ||
|
|
"accounts/fireworks/models/llama-v3p1-8b-instruct",
|
|
apiKey: fireworksApiKey,
|
|
configuration: {
|
|
baseURL: "https://api.fireworks.ai/inference/v1",
|
|
},
|
|
streamUsage: false,
|
|
});
|
|
Object.defineProperty(this, "lc_serializable", {
|
|
enumerable: true,
|
|
configurable: true,
|
|
writable: true,
|
|
value: true
|
|
});
|
|
Object.defineProperty(this, "fireworksApiKey", {
|
|
enumerable: true,
|
|
configurable: true,
|
|
writable: true,
|
|
value: void 0
|
|
});
|
|
Object.defineProperty(this, "apiKey", {
|
|
enumerable: true,
|
|
configurable: true,
|
|
writable: true,
|
|
value: void 0
|
|
});
|
|
this.fireworksApiKey = fireworksApiKey;
|
|
this.apiKey = fireworksApiKey;
|
|
}
|
|
getLsParams(options) {
|
|
const params = super.getLsParams(options);
|
|
params.ls_provider = "fireworks";
|
|
return params;
|
|
}
|
|
toJSON() {
|
|
const result = super.toJSON();
|
|
if ("kwargs" in result &&
|
|
typeof result.kwargs === "object" &&
|
|
result.kwargs != null) {
|
|
delete result.kwargs.openai_api_key;
|
|
delete result.kwargs.configuration;
|
|
}
|
|
return result;
|
|
}
|
|
/**
|
|
* Calls the Fireworks API with retry logic in case of failures.
|
|
* @param request The request to send to the Fireworks API.
|
|
* @param options Optional configuration for the API call.
|
|
* @returns The response from the Fireworks API.
|
|
*/
|
|
async completionWithRetry(request, options) {
|
|
delete request.frequency_penalty;
|
|
delete request.presence_penalty;
|
|
delete request.logit_bias;
|
|
delete request.functions;
|
|
if (request.stream === true) {
|
|
return super.completionWithRetry(request, options);
|
|
}
|
|
return super.completionWithRetry(request, options);
|
|
}
|
|
}
|