IntelliNode v1.4.2

New Changes 🌟

• Add a chat context function to manage the relevant messages for chatbots.
• Update the module to support next.js integration.

Code Examples 💻

• Check example of next.js integration with a modern chatbot interface.

• Try the below Code to manage the chatbot history strategy.

const { ChatContext } = require('intellinode');
const context = new ChatContext(apiKey);

# Parameters
const userMessage = "Hello";
const historyMessages = ["Good morning", "Dinner time", "How can I help you?", "Hello"];
const n = 3;

# Execute the filter 
const resultContext = await context.getStringContext(userMessage, historyMessages, n);

This code manages the chatbot window size limitation by returning the relevant messages for the user input. Define a window size (n), and the function will filter the last two messages in addition to the relevant ones from the history fitting the window size.

Full chat context sample available here

Contributors 👏

• @Eytyy fixed next.js integration.
• @Barqawiz added the chat context functionality.

IntelliNode v1.3.9

New Changes 🌟

• ChatGPT Streaming: Added support for ChatGPT streaming, allowing users to receive a stream of responses instead of waiting for the entire conversation to complete.
• OpenAI Whisper Support: Added support for speech-to-text functionality.

Code Examples 💻

• Install the module: npm i intellinode
• Code example for ChatGPT streaming: test_chatbot.js.
• Chatbot documentation with streaming instructions.

Contributors 👏

• @Barqawiz updated the chatbot to support streaming.
• @pushoo-sharma added whisper integration.

IntelliNode v1.3.7

New Changes 🌟

• Support Llama Code and Python Models: now you can utilize Llama code and python-based models in the intelliNode module.
• Large Data Memory Semantic Search: introducing enhanced semantic search with large data using pagination.

Code Examples 💻

• Install the module: npm i intellinode
• The llama coder sample: test_llama_chatbot.
• The large data semantic search using SemanticSearchPaging: test_semantic_search_pagination.

IntelliNode v1.3.0

v1.3.0

New Changes 🌟

• Introduced the Large Learning Model (LLM) Evaluator: a quick comparison among multiple models' results based on your target answers. This feature simplifies the assessing of LLMs such as LaMMA V2 and GPT4, providing valuable insights into model performance.
• Prompt Engineering Feature: an efficient way to automate the process of creating new prompts.

Code Examples 💻

LLM Evaluator

const { LLMEvaluation } = require('intellinode');

// create the LLMEvaluation with the desired embedding for unified comparison.
const llmEvaluation = new LLMEvaluation(openaiKey, 'openai');

const results = await llmEvaluation.compareModels(inputString, targetAnswers, providerSets)

Check the model evaluation documentation.

Prompt Engineering

Import the Prompt class:

const Prompt = require('intellinode');

Generate a new prompt using an existing model:

const promptTemp = await Prompt.fromChatGPT("generate a fantasy image with ninja jumping across buildings", openaiApiKey);

Automatically fill and format a prompt with dynamic data:

const promptTemp = await Prompt.fromText("information retrieval about ${reference} text with ${query} user input");
const inputString = promptTemp.format({reference: '<reference_text>', query: '<user_query>'})

IntelliNode v1.2.3 🦙

v1.2.3

Update Chatbot with Llama V2 Support 🌟

IntelliNode library now supports Llama 🦙 v2 chat models, enhancing the capabilities of the chatbot function even further.

The Llama model available with two options:

1. Replica API Integration: a seamless and fast implementation requiring only an API key. Users can integrate Llama v2 into their existing apps with minimal configuration using intelli-node module.

2. AWS SageMaker Integration: a more advanced and flexible integration adaptability for enterprise-level projects.

Code Examples 💻

The documentation on how to switch the chatbot between ChatGPT and LLama can be found in the IntelliNode chatbot wiki.

IntelliNode v1.1.0

v1.1.0

New Changes 🌟

• Upgraded the chatbot to support functions call which empowers the automation.
• Added an automation sample for AWS S3.
• Openai updated to support the organization as an optional parameter.

Code Examples 💻

Automation

import:

const { Chatbot, ChatGPTInput, ChatGPTMessage, FunctionModelInput } = require('intellinode');

call:

// define the functions details
const functions_desc = [
  new FunctionModelInput('list_buckets', 'List all available S3 buckets')
];

// prepare the input
const systemMsg = "Don't make assumptions about what values to plug into functions. Ask for clarification if a user request is ambiguous.";
const input = new ChatGPTInput(systemMsg, {model: "gpt-3.5-turbo-0613"});
input.addMessage(new ChatGPTMessage("List my s3 buckets.", "user"));

// call the chatbot
const bot = new Chatbot(apiKey, SupportedChatModels.OPENAI);
const responses = await bot.chat(input, functions_desc.map(f => f.getFunctionModelInput()));

parse the output:

// if the response type is an object, the model response a function name and arguments to call.
// otherwise, the model returned a text response for clarification.
responses.forEach((response) => {
  if (typeof response === "object") {
    console.log("- Function call: ", JSON.stringify(response.function_call, null, 2));
  } else {
    console.log("- " + response);
  }
});

Check the full automation example under the samples directory.

IntelliNode v1.0.0

v1.0.0

🎉 The first stable release of IntelliNode, after testing with hundreds of users via the online playground show.intellinode.ai

New Changes 🌟

• Update the Gen function to support the new 16k context version of gpt-3.5-turbo.
• Security enhancements.
• Add test runner to validate the code before publishing to npm repository.

IntelliNode v0.0.31

v0.0.31

New Changes 🌟

• Support Azure OpenAI backend.
• Add an optional openai proxy URL to bypass regional restrictions (request by the users).
• Improve the Gen function by accepting backend engine to switch between models in one line code.
• Enhance the csv to html charts function.

Code Examples 💻

Openai Azure Service

1. Set Azure backend as proxy

const { ProxyHelper } = require('intellinode');
ProxyHelper.getInstance().setAzureOpenai(resourceName);

2. Call any openai function

const { Gen } = require('intellinode');
// generate blog post
result = await Gen.get_blog_post(prompt, openaiApiKey, provider='openai');

Gen content with multi backend example

• Generate marketing content using openai

marketingDesc = await Gen.get_marketing_desc(prompt, openaiApiKey);

• Or using cohere

marketingDesc = await Gen.get_marketing_desc(prompt, cohereApiKey, provider='cohere');

Gen image with multi backend example

• Generate prompt description using chatGPT and the image using stable diffusion:

image = await Gen.generate_image_from_desc(prompt, openaiApiKey, stabilityApiKey, true);

• or generate the image using DALL·E:

image = await Gen.generate_image_from_desc(prompt, openaiApiKey, openaiApiKey, true, provider='openai');

IntelliNode v0.0.21

v0.0.21

New Changes 🌟

• Introducing a new Gen function: Generate charts directly from CSV data.
• Enhanced the HTML generation: we have further enhanced the Gen function to generate HTML pages based on user prompts.

Code Examples 💻

• import:

const { Gen } = require('intellinode');

Generate charts

const csv_str_data = '<your csv as string>'
const topic = "<the csv topic>";

const htmlCode = await Gen.generate_dashboard(csv_str_data, topic, openaiKey, num_graphs=2);

Output example:

Generate html files

text = 'a registration page with flat modern theme.'
await Gen.save_html_page(text, folder='tempFolder', file_name='register_page', openaiKey='<key>');

Output example:

Note that Gen function in beta, the output generated may not always meet your expectations.

For more code examples, check the sample.

IntelliNode v0.0.19

v0.0.19

New Features 🌟

• Add a new Gen function to generate and save html pages from users prompt.

Code Examples 💻

• import:

const { Gen } = require('intellinode');

• execute the one line command to generate html:

text = 'a registration page with flat modern theme.'
await Gen.save_html_page(text, folder='tempFolder', file_name='register_page', openaiKey='<key>');

Output example:

• other existing functions

// one line to generate blog post
const blogPost = await Gen.get_blog_post(prompt, openaiApiKey);

For more code examples, check the sample.