WeAviate fetch similar vector #5656
Replies: 1 comment 5 replies
-
|
To properly query Weaviate to retrieve similar embeddings and send them to OpenAI, you can follow these steps:
Here is an example code snippet that demonstrates these steps: import weaviate, { ApiKey } from "weaviate-ts-client";
import { OpenAIEmbeddings, OpenAI } from "@langchain/openai";
import { SelfQueryRetriever } from "langchain/retrievers/self_query";
import { WeaviateStore, WeaviateTranslator } from "@langchain/weaviate";
import { Document } from "@langchain/core/documents";
import { AttributeInfo } from "langchain/chains/query_constructor";
// Create documents
const docs = [
new Document({
pageContent: "A bunch of scientists bring back dinosaurs and mayhem breaks loose",
metadata: { year: 1993, rating: 7.7, genre: "science fiction" },
}),
new Document({
pageContent: "Leo DiCaprio gets lost in a dream within a dream within a dream within a ...",
metadata: { year: 2010, director: "Christopher Nolan", rating: 8.2 },
}),
new Document({
pageContent: "A psychologist / detective gets lost in a series of dreams within dreams within dreams and Inception reused the idea",
metadata: { year: 2006, director: "Satoshi Kon", rating: 8.6 },
}),
new Document({
pageContent: "A bunch of normal-sized women are supremely wholesome and some men pine after them",
metadata: { year: 9.3, director: "Greta Gerwig", rating: 8.3 },
}),
new Document({
pageContent: "Toys come alive and have a blast doing so",
metadata: { year: 1995, genre: "animated" },
}),
new Document({
pageContent: "Three men walk into the Zone, three men walk out of the Zone",
metadata: {
year: 1979,
director: "Andrei Tarkovsky",
genre: "science fiction",
rating: 9.9,
},
}),
];
// Define attributes
const attributeInfo: AttributeInfo[] = [
{ name: "genre", description: "The genre of the movie", type: "string or array of strings" },
{ name: "year", description: "The year the movie was released", type: "number" },
{ name: "director", description: "The director of the movie", type: "string" },
{ name: "rating", description: "The rating of the movie (1-10)", type: "number" },
{ name: "length", description: "The length of the movie in minutes", type: "number" },
];
// Instantiate vector store
const embeddings = new OpenAIEmbeddings();
const llm = new OpenAI({ modelName: "gpt-3.5-turbo" });
const client = (weaviate as any).client({
scheme: process.env.WEAVIATE_SCHEME || "https",
host: process.env.WEAVIATE_HOST || "localhost",
apiKey: new ApiKey(process.env.WEAVIATE_API_KEY || "default"),
});
const vectorStore = await WeaviateStore.fromDocuments(docs, embeddings, {
client,
indexName: "Test",
textKey: "text",
metadataKeys: ["year", "director", "rating", "genre"],
});
// Create SelfQueryRetriever
const selfQueryRetriever = SelfQueryRetriever.fromLLM({
llm,
vectorStore,
documentContents: "Brief summary of a movie",
attributeInfo,
structuredQueryTranslator: new WeaviateTranslator(),
});
// Query the vector store
const query1 = await selfQueryRetriever.getRelevantDocuments("Which movies are rated higher than 8.5?");
const query2 = await selfQueryRetriever.getRelevantDocuments("Which movies are directed by Greta Gerwig?");
console.log(query1, query2);This code sets up the documents, defines the attributes, initializes the vector store, creates a |
Beta Was this translation helpful? Give feedback.
-
|
No I want to do it for swagger file that I have stored in weaviate I have again made question embedding and need to get nearby embedding |
Beta Was this translation helpful? Give feedback.
-
|
To query Weaviate for nearby embeddings of your Swagger file and send them to OpenAI, follow these steps:
Here is an example code snippet: Step 1: Query Weaviate for Nearby Embeddingsimport weaviate, { ApiKey } from "weaviate-ts-client";
import { WeaviateStore } from "@langchain/weaviate";
import { OpenAIEmbeddings } from "@langchain/openai";
async function getNearbyEmbeddings(swaggerEmbedding: number[]) {
const client = (weaviate as any).client({
scheme: process.env.WEAVIATE_SCHEME || "https",
host: process.env.WEAVIATE_HOST || "localhost",
apiKey: new ApiKey(process.env.WEAVIATE_API_KEY || "default"),
});
const store = await WeaviateStore.fromExistingIndex(new OpenAIEmbeddings(), {
client,
indexName: "YourClassName",
metadataKeys: ["_additional { vector }"],
});
const results = await store.similaritySearch(swaggerEmbedding, 10);
return results.map((item: any) => item.metadata.vector);
}Step 2: Send Embeddings to OpenAIimport { OpenAIEmbeddings } from "@langchain/openai";
async function sendEmbeddingsToOpenAI(embeddings: number[][]) {
const openAIEmbeddings = new OpenAIEmbeddings({
apiKey: 'your-openai-api-key',
});
const results = [];
for (const embedding of embeddings) {
const result = await openAIEmbeddings.embedQuery(embedding);
results.push(result);
}
return results;
}
// Example usage
(async () => {
const swaggerEmbedding = [/* your Swagger file embedding */];
const nearbyEmbeddings = await getNearbyEmbeddings(swaggerEmbedding);
const openAIResults = await sendEmbeddingsToOpenAI(nearbyEmbeddings);
console.log(openAIResults);
})();This code first queries Weaviate to get the nearby embeddings for the Swagger file and then sends those embeddings to OpenAI using the |
Beta Was this translation helpful? Give feedback.
-
|
I have entered this code import * as fs from "fs"; const OPENAI_API_KEY = process.env.OPENAI_API_KEY; You should build the API URLs in order to get a response that is as short as possible, while still getting the necessary information to answer the question. Pay attention to deliberately exclude any unnecessary pieces of data in the API call. // let finalData = await model.generate([apiUrlPrompt]); async function splitTextIntoChunks(text) { async function generateEmbeddings(textChunks) { } async function getNearbyEmbeddings(queryEmbedding) { } async function main() { } main(); But getting this error Error: TypeError: text.replace is not a function |
Beta Was this translation helpful? Give feedback.
-
|
The error Here is the corrected code with necessary type checks and adjustments:
Updated Codeimport * as fs from "fs";
import { config } from 'dotenv';
import { createHash } from 'crypto';
import { RecursiveCharacterTextSplitter } from "langchain/text_splitter";
import { OpenAIEmbeddings } from "@langchain/openai";
import weaviate, { ApiKey } from "weaviate-ts-client";
import { WeaviateStore } from "@langchain/weaviate";
import { OpenAI } from "@langchain/openai";
import { PromptTemplate } from "@langchain/core/prompts";
config();
const OPENAI_API_KEY = process.env.OPENAI_API_KEY;
const model = new OpenAI({ temperature: 0, apiKey: OPENAI_API_KEY });
const API_URL_PROMPT_TEMPLATE1 = `You are given the below the embedding fo some swagger Documentation:
{api_docs}
Using this embeddings, find the relevant API URLs there can be multiple url so give all the url step by step give the method also like get, post , put , delete and give the answer in step1 , step 2 like this.
You should build the API URLs in order to get a response that is as short as possible, while still getting the necessary information to answer the question. Pay attention to deliberately exclude any unnecessary pieces of data in the API call.
Question:{question}
API url:`;
const API_URL_PROMPT_TEMPLATE = new PromptTemplate({
inputVariables: ["api_docs", "question"],
template: API_URL_PROMPT_TEMPLATE1,
});
function generateHash(text) {
return createHash('sha256').update(text).digest('hex');
}
async function splitTextIntoChunks(text) {
const splitter = new RecursiveCharacterTextSplitter({ chunkSize: 1000, chunkOverlap: 200 });
const documents = await splitter.createDocuments([text]);
return documents.map(doc => doc.pageContent);
}
async function generateEmbeddings(text) {
if (typeof text !== 'string') {
throw new TypeError('Input text must be a string');
}
const embeddings = new OpenAIEmbeddings({ apiKey: process.env.OPENAI_API_KEY });
return await embeddings.embedQuery(text);
}
async function getNearbyEmbeddings(queryEmbedding) {
const client = (weaviate).client({
scheme: process.env.WEAVIATE_SCHEME || "https",
host: "swagger-as4wyy7e.weaviate.network",
apiKey: new ApiKey('yEhKKW4MLmbvVJT2yw7mlTYutuz5zl22U2p1'),
});
const store = await WeaviateStore.fromExistingIndex(new OpenAIEmbeddings(), {
client,
indexName: "Test",
textKey: "text",
metadataKeys: ["hash"],
});
const results = await store.similaritySearch(queryEmbedding, 10);
return results.map((item) => item.metadata.vector);
}
async function main() {
try {
const specFilePath = './swagger.json';
const jsonData = fs.readFileSync(specFilePath, 'utf8');
console.log("This is json data:", jsonData);
const hash = generateHash(jsonData);
const textChunks = await splitTextIntoChunks(jsonData);
await WeaviateStore.fromTexts(
textChunks,
textChunks.map(chunk => ({ hash })),
new OpenAIEmbeddings(),
{
client,
indexName: "Test",
textKey: "text",
metadataKeys: ["hash"],
}
);
const query = "Can you give the api url for whitelist investor and then fetch all wallet?";
const queryEmbedding = await generateEmbeddings(query);
console.log("This is query embedding", queryEmbedding);
const similarEmbeddings = await getNearbyEmbeddings(queryEmbedding);
console.log("This is similar embedding", similarEmbeddings);
const apiUrlPrompt = await API_URL_PROMPT_TEMPLATE.format({
api_docs: similarEmbeddings,
question: query,
});
const apiUrlResponse = await model.generate([apiUrlPrompt]);
console.log("API URL Response:", JSON.stringify(apiUrlResponse, null, 2));
const apiUrl = apiUrlResponse.generations[0][0]?.text;
console.log("Generated API URL:", apiUrl);
} catch (error) {
console.error('Error:', error);
}
}
main();Key Changes:
This should resolve the |
Beta Was this translation helpful? Give feedback.
-
|
OK so in getNearByEmbedding we need to pass only query it will automatically convert in embedding and then find the reultant near embedding |
Beta Was this translation helpful? Give feedback.
-
Checked other resources
Commit to Help
Example Code
Description
In this code I have stored my data in weaviate now I want to fetch the similar embedding I dont know how to query it is not giving similar embedding how to get it so I can send it to openAI
System Info
all good
Beta Was this translation helpful? Give feedback.
All reactions