Merge pull request #7 from upstash/dx-973-upstash-llm-client-for-rag-…

…chat

DX-973: Upstash LLM Client

src/upstash-llm-client.test.ts

@@ -0,0 +1,127 @@
+/* eslint-disable unicorn/numeric-separators-style */
+/* eslint-disable @typescript-eslint/no-non-null-assertion */
+import type { AIMessage } from "@langchain/core/messages";
+import { Redis } from "@upstash/redis";
+import { Index } from "@upstash/vector";
+import type { StreamingTextResponse } from "ai";
+import { afterAll, beforeAll, describe, expect, test } from "bun:test";
+
+import { RAGChat } from "./rag-chat";
+import { awaitUntilIndexed } from "./test-utils";
+import { UpstashLLMClient } from "./upstash-llm-client";
+
+describe("RAG Chat with Upstash LLM Client", () => {
+  const vector = new Index({
+    token: process.env.UPSTASH_VECTOR_REST_TOKEN!,
+    url: process.env.UPSTASH_VECTOR_REST_URL!,
+  });
+  afterAll(async () => await vector.reset());
+
+  describe("meta-llama/Meta-Llama-3-8B-Instruct", () => {
+    const client = new UpstashLLMClient({
+      model: "meta-llama/Meta-Llama-3-8B-Instruct",
+      apiKey: process.env.UPSTASH_LLM_REST_TOKEN!,
+      streaming: true,
+    });
+
+    const ragChat = new RAGChat({
+      model: client,
+      vector,
+      redis: new Redis({
+        token: process.env.UPSTASH_REDIS_REST_TOKEN!,
+        url: process.env.UPSTASH_REDIS_REST_URL!,
+      }),
+    });
+
+    beforeAll(async () => {
+      await ragChat.addContext({
+        dataType: "text",
+        data: "Paris, the capital of France, is renowned for its iconic landmark, the Eiffel Tower, which was completed in 1889 and stands at 330 meters tall.",
+      });
+      await awaitUntilIndexed(vector);
+    });
+
+    test(
+      "should get result without streaming",
+      async () => {
+        const result = (await ragChat.chat(
+          "What year was the construction of the Eiffel Tower completed, and what is its height?",
+          { stream: false }
+        )) as AIMessage;
+
+        expect(result.content).toContain("330");
+      },
+      { timeout: 10000 }
+    );
+
+    test(
+      "should get result with streaming",
+      async () => {
+        const result = (await ragChat.chat(
+          "Which famous artworks can be found in the Louvre Museum?",
+          {
+            stream: true,
+          }
+        )) as StreamingTextResponse;
+
+        expect(result).toBeTruthy();
+      },
+      { timeout: 10000 }
+    );
+  });
+
+  describe("mistralai/Mistral-7B-Instruct-v0.2", () => {
+    const client = new UpstashLLMClient({
+      model: "mistralai/Mistral-7B-Instruct-v0.2",
+      apiKey: process.env.UPSTASH_LLM_REST_TOKEN!,
+      streaming: true,
+    });
+
+    const ragChat = new RAGChat({
+      model: client,
+      vector,
+      redis: new Redis({
+        token: process.env.UPSTASH_REDIS_REST_TOKEN!,
+        url: process.env.UPSTASH_REDIS_REST_URL!,
+      }),
+    });
+
+    beforeAll(async () => {
+      await ragChat.addContext({
+        dataType: "text",
+        data: "Paris, the capital of France, is renowned for its iconic landmark, the Eiffel Tower, which was completed in 1889 and stands at 330 meters tall.",
+      });
+      await awaitUntilIndexed(vector);
+    });
+
+    // afterAll(async () => await vector.reset());
+
+    test(
+      "should get result without streaming",
+      async () => {
+        const result = (await ragChat.chat(
+          "What year was the construction of the Eiffel Tower completed, and what is its height?",
+          { stream: false }
+        )) as AIMessage;
+
+        expect(result.content).toContain("330");
+      },
+      { timeout: 10000 }
+    );
+
+    test(
+      "should get result with streaming",
+      async () => {
+        const result = (await ragChat.chat(
+          "Which famous artworks can be found in the Louvre Museum?",
+          {
+            stream: true,
+          }
+        )) as StreamingTextResponse;
+
+        expect(result).toBeTruthy();
+      },
+      { timeout: 10000 }
+    );
+  });
+});

src/upstash-llm-client.ts

@@ -0,0 +1,80 @@
+import { ChatOpenAI } from "@langchain/openai";
+import { type BaseMessage } from "@langchain/core/messages";
+import { type ChatGeneration } from "@langchain/core/outputs";
+
+export type Model = "mistralai/Mistral-7B-Instruct-v0.2" | "meta-llama/Meta-Llama-3-8B-Instruct";
+
+export type UpstashLLMClientConfig = {
+  model: Model;
+  apiKey: string;
+  streaming: boolean;
+  maxTokens?: number;
+  stop?: string[];
+  topP?: number;
+  temperature?: number;
+  frequencyPenalty?: number;
+  presencePenalty?: number;
+  n?: number;
+  logitBias?: Record<string, number>;
+  logProbs?: number;
+  topLogprobs?: number;
+};
+
+export class UpstashLLMClient extends ChatOpenAI {
+  modelName: Model;
+  apiKey: string;
+  maxTokens?: number;
+  stop?: string[];
+  temperature = 1;
+  n = 1;
+  streaming: boolean;
+  topP = 1;
+  frequencyPenalty = 0;
+  presencePenalty = 0;
+  logitBias?: Record<string, number>;
+  logProbs?: number;
+  topLogprobs?: number;
+
+  constructor(config: UpstashLLMClientConfig) {
+    super(
+      {
+        modelName: config.model,
+        apiKey: config.apiKey,
+        maxTokens: config.maxTokens,
+        streaming: config.streaming,
+        topP: config.topP,
+        temperature: config.temperature,
+        n: config.n,
+        frequencyPenalty: config.frequencyPenalty,
+        presencePenalty: config.presencePenalty,
+        logitBias: config.logitBias,
+        topLogprobs: config.topLogprobs,
+        stop: config.stop,
+      },
+      {
+        baseURL: "https://qstash.upstash.io/llm/v1",
+      }
+    );
+
+    this.modelName = config.model;
+    this.apiKey = config.apiKey;
+    this.maxTokens = config.maxTokens;
+    this.streaming = config.streaming;
+    this.logitBias = config.logitBias;
+    this.topLogprobs = config.topLogprobs;
+    this.stop = config.stop;
+
+    // @ts-expect-error This is overriding the method
+    this.getNumTokensFromGenerations = (_generations: ChatGeneration[]): Promise<number> => {
+      return Promise.resolve(0);
+    };
+
+    this.getNumTokensFromMessages = (
+      _messages: BaseMessage[]
+    ): Promise<{ totalCount: number; countPerMessage: number[] }> => {
+      return new Promise((resolve, _) => {
+        resolve({ totalCount: 0, countPerMessage: [0] });
+      });
+    };
+  }
+}