Instant AI Model Tuning : Leveraging HNSW Vector with Firebase Genkit for Retrieval-Augmented Generation | by R Surahutomo Aziz Pradana | May, 2024

Now let’s comply with to review additional how we are going to assemble such an AI Reply!

Circumstances

Sooner than placing within the plugin, assure you’ve got bought the subsequent put in:

• Node.js (mannequin 12 or bigger)
• npm (comes with Node.js)
• TypeScript (arrange globally by npm: npm arrange -g typescript)
• Genkit (arrange globally by npm: npm arrange -g genkit)

Very very first thing first, provoke Genkit mission with

genkit init

adjust to the instructions here.

After getting the genkit mission put in, ensure that the mission prepared properly, you can try first by

genkit start

If it runs properly and open the Genkit UI in a browser, then you definitely could be good to go!

Placing within the HNSW plugin

To place within the Genkit HNSW plugin, run the subsequent command:

npm arrange genkitx-hnsw

We will be using 2 Genkit Plugins proper right here.

1. HNSW Indexer plugin
2. HNSW Retriever plugin

1.HNSW Indexer Plugin

The HNSW Indexer plugin helps create a vector index out of your info, which may be utilized as a info reference for the HNSW Retriever.

Data Preparation

Put collectively your info or paperwork, for instance, restaurant info, in a loyal folder.

Registering the HNSW Indexer Plugin

Import the plugin into your Genkit mission:

uncover genkit.config.ts file in your mission, typically /root/src/genkit.config.ts.

Then import the plugin into the file.

import { hnswIndexer } from "genkitx-hnsw";

// 
export default configureGenkit({
plugins: [
hnswIndexer({ apiKey: "GOOGLE_API_KEY" })
]
});

Working the Indexer

1. Open the Genkit UI and select the registered HNSW Indexer plugin.
2. Execute the stream with the required parameters:

• dataPath: Path to your info and paperwork.
• indexOutputPath: Desired output path for the generated vector retailer index.

Vector Retailer Index Finish outcome

The HNSW vector retailer shall be saved throughout the specified output path, ready for use with the HNSW Retriever plugin.

The HNSW Retriever plugin processes prompts with the Gemini LLM Model, enriched with additional specific information from the HNSW Vector index.

Registering the HNSW Retriever Plugin

Import the required plugins into your Genkit mission:

import { googleAI } from "@genkit-ai/googleai";
import { hnswRetriever } from "genkitx-hnsw";

export default configureGenkit({
plugins: [
googleAI(),
hnswRetriever({ apiKey: "GOOGLE_API_KEY" })
]
});

Working the Retriever

1. Open the Genkit UI and select the HNSW Retriever plugin.
2. Execute the stream with the required parameters:

• quick: Your enter query for the AI.
• indexPath: Path to the vector index file generated by the HNSW Indexer plugin.

Occasion Fast

To ask regarding the value itemizing of a restaurant in Surabaya Metropolis:

quick: "What is the value itemizing of my restaurant in Surabaya Metropolis?"
indexPath: "/path/to/your/vector/index"

The mix of HNSW Vector index with Genkit significantly enhances the capabilities of Generative AI fashions by providing enriched context and specific info.

This technique not solely improves the accuracy of AI responses however moreover simplifies the strategy of knowledge integration, making it a powerful system for quite a few functions.

By following the steps outlined on this text, you can efficiently leverage the HNSW Vector index to assemble additional intelligent and context-aware AI applications in a very fast time like instantly!

Hope this helps and see you throughout the subsequent one!