Node&#s GoogleGenerativeAI: Incorporating AI Technology In javaScript

Introduction

In recent years, the field of artificial intelligence has made great progress. With the increasing popularity of artificial intelligence, developers must find a way to integrate AI into their applications. Gemini provides a convenient AI development approach for JavaScript developers through Node's Googlegene asynchronous Pack. Developers can access the Gemini model developed by Google DeepMind to create exciting features with AI. Python or Go users can use other software packages, Gemini also provides RESTFUL API. This article will discuss the improvements brought by the latest model of Gemini, and how to use Node's Googlegenerativeai to wrap into the door.

The main progress

A major improvement of Gemini 1.5 Flash model is the number of context marks in a single request. In the past, such models were limited by the number of texts or marks that can be processed at a time. The generating model created in the past few years can only process 8,000 marks at a time. Although this number has improved with the advancement of artificial intelligence technology, it is still a limited factor. Today, Gemini 1.5 Flash can handle up to 1 million marks at a time. Professional version (Gemini 1.5 Pro) can handle up to 2 million labels. This allows Gemini to process a lot of information at a time while maintaining a very high accuracy. You can read more information about Gemini's progress and significance in the field of artificial intelligence.

Getting Started

To use the Googlegeneramedai package, you first need to create a Gemini API key. This is a fast and simple process.

Go to Google Ai Studio

Click the "Get API Key" button in the upper left corner

1. Click the "Create API Key" button
After visiting the API key, you need to install the software package with Node.
3. After completing all these operations, you can start using AI for development!

Set

npm install @google/generative-ai

Guide the software into the file you want to use it.

Create a GooglegeNerabAIVEAI instance and pass in your API key at the same time.

Use the getGENERATIVEMDEL method and pass it into the model object you want to use. There are multiple models available. This example uses the Gemini 1.5 Flash model. Gemini model

import { GoogleGenerativeAI } from '@google/generative-ai';
// 或
const { GoogleGenerativeAI } = require('@google/generative-ai');

After setting the model, you can use AI to generate text, respond images, extract information from video, and so on.

const genAI = new GoogleGenerativeAI('YOUR_API_KEY');

Configuration and system instructions

const model = genAI.getGenerativeModel({ model: 'gemini-1.5-flash' });

You can choose to provide configuration and system instructions for the model. Configure the GenerationConfig attribute in the GenerateContent method call. Some configuration options include:

• ResponseSchema: The output mode of the generated text
• CandidateCount: (integer) The number of responses to return
• Temperator: (Digital) The randomness of the output controls the output

See more generatingconfig attributes here. Provide system instructions to help improve response by providing more contexts for AI. In addition, the model will generate more customized responses and can better meet the needs of users. Provide system instructions when initialized models.

import { GoogleGenerativeAI } from '@google/generative-ai';
// 或
const { GoogleGenerativeAI } = require('@google/generative-ai');

Text generation

You can use multiple methods to use the software to form a text. The easiest way is to provide text only for the model, but there are more exciting and complex methods to generate text. You can provide images and text for the model so that AI responds to the image. This is a simple example of a request that uses a text to generate response. The model settings are not included in this code block, but it is still part of the code.

const genAI = new GoogleGenerativeAI('YOUR_API_KEY');

The prompt string is passed to the model of the model. After returning to the response, you can access the response in the Text method of the response property. The result of this response is: "The surface of the moon is covered with a fine dust called the weathering layer, which is formed by the impact of billions of years. This dust is very thin and will stick to everything. Is it cool to make a challenge with the moon equipment? This is a super simple example, but there are more possibilities.

Text flow and chat

The model is waiting to generate the entire response text before returning the response. Obviously, right? If you don't want to wait for the entire response, you can use text flow to get faster response by not waiting for the whole result. This can be implemented using the StreamGenerateContent method. The following is an example in the Gemini API document.

The software package also provides the function of tracking dialogue. "Allow users to find the answer step by step", which helps users solve multiple steps. This is a relatively advanced feature of Gemini API. For more information about creating chat and other text generation functions, read the Gemini API documentation.

const model = genAI.getGenerativeModel({ model: 'gemini-1.5-flash' });

Conclusion

Googlegenerativeai package enables JavaScript developers to easily integrate its applications into AI technology. The software package has a variety of functions in the generation of AI, including text, videos and images. Gemini's ability to process a large number of texts at a time is a major development generated by AI. With Node's Googlegeneramedai, developers can include advanced AI technologies in their projects in a simpler way.

Source

npm Deepmind Gemini long context

The above is the detailed content of Node&#s GoogleGenerativeAI: Incorporating AI Technology In javaScript. For more information, please follow other related articles on the PHP Chinese website!