Community Engagement: Public Meetings

Join us for public ZOOM meetings. Stay informed, voice your opinions.�Public Calendar

Agenda

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User

​

Hands-On Testing - Developer�

Network , Social Media Links

Innovative Features

Moderate Q and A

AGENDA

Welcome

Introduction to AIFAQ Chatbot - Supratick Mondal

Live Demonstration - End User

​

Hands-On Testing - Developer�

Network , Social Media Links

Innovative Features

Moderate Q and A

Development and Functionalities:

GOAL �To assist users by providing a conversational AI tool that efficiently answers context-specific questions, reducing the need to sift through extensive documentation.

​

We aim to support the users by simplifying workflows by avoiding the need to wade through oceans of documents to find information they are looking for.

​

​

Development and Functionalities:

FUNCTIONALITY

The prototype implements an open-source AI chatbot that responds to HTTP requests using a RESTful API.

​

It utilizes Retrieval Augmented Generation (RAG) to enhance the chatbot's answers by incorporating knowledge from external databases. The system supports data ingestion to create a vector database from sources like online software guides and GitHub documentation.

​

​

AIFAQ Github Repo

Development and Functionalities:

INTEGRATION (Hyperledger)

The system is an open-source Python project that integrates with external knowledge databases, such as vector databases, to improve AI responses.

​

It currently uses the HuggingFace Zephyr-7b-beta model, with plans to explore additional open-source models. A user interface (UI) module is also planned for future development to complement the existing HTTP request-based.

Data Ingestion Workflow:

• Documents from ReadTheDocs are loaded, split, and chunked.

​

• These text chunks are then processed by the HuggingFace Embedding Model, which converts them into vector embeddings.

​

• The resulting vectorized chunks (embeddings) and their corresponding text chunks are combined and stored in ChromaDB.

​

​

• ChromaDB manages these embeddings to facilitate efficient retrieval and similarity searches for later queries.

​

Chat Workflow:

• When a user submits a text query, it is sent to the HuggingFace model, which generates vector embeddings of the query.

​

• These embeddings are then sent to ChromaDB, a vector database that performs a similarity search to find relevant text chunks from previously stored embeddings.

​

• The query and these retrieved text chunks are then forwarded to the Zephyr-7B Beta LLM, which generates a response based on the combined input.

​

• Finally, the response is passed through LangChain and FastAPI services, which format and deliver it back to the user.

​

• This process ensures efficient, contextually accurate responses by leveraging advanced models and vector search capabilities.

​

User Interface/ Frontend

AGENDA

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User - Xitong (Jacqueline) Zhang

​

Hands-On Testing - Developer�

Network , Social Media Links

Innovative Features

​

Moderate Q and A

Live Demonstration - User Interaction

Setup

• Chat Bot
• Associated Documents
• Knowledge Base

Interaction

• Real-time Responses to Queries

Response Analysis

• Response
• Source

AGENDA

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User

​

Hands-On Testing - Developer�

Network , Social Media Links

Innovative Features - Madugula Jayaram

Moderate Q and A

Innovative Features

LLM

Large Language Model + RAG �Supporting Documentation

Open Source Repo

OPEN SOURCE�Preserve data privacy, reduce costs, high flexibility

QLoRA� Memory Usage Reduction

Quantized Low-Rank technique Supporting Documentation

Innovative Features

LLM

Large Language Model + RAG

Innovative Features

OPEN SOURCE Preserve data privacy, reduce costs, high flexibility

Innovative Features

QLoRA Memory usage reduction, Quantized Low-Rank technique

AGENDA

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User

​

Hands-On / Testing - Developer - Gianluca Capuzzi�

Network , Social Media Links

Innovative Features

Moderate Q and A

Help Us Test

Testing

Install the AIFAQ Chatbot:� YouTube hands-on guide

Feedback

Provide insights for improvement:

Feedback Form

Engage

Join the conversation:

Discord Channel

AGENDA

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User

​

Hands-On / Testing - Developer�

Network , Social Media Links - Shreya Sahay and Anshika Vashistha

​

Innovative Features

Moderate Q and A

Networking Opportunities

1

Connect

Meet like-minded individuals in AI and blockchain

2

Exchange

Share ideas and explore collaboration opportunities

3

Build

Contribute to a community of innovators

AGENDA

Welcome

Introduction to AIFAQ Chatbot

Live Demonstration - End User

​

Hands-On / Testing - Developer

Network , Social Media Links

Innovative Features

Moderate Q and A - Arunima Chaudhury

Q&A

​

Discord Link https://discord.gg/hyperledger

​

Website https://hyperledger.org/labs/aifaq

​

GitHub https://github.com/hyperledger-labs/aifaq

Future improvements

Cloud installation

Allow the prototype installation on a Cloud Server: React frontend

Further use cases

Expert systems, Dataset generation

Small AI Model

Train a smaller model: cost reduction and performance improvement

Hands-On Testing Step By Step

SETUP THE AIFAQ Bot

Step 1: Sign Up to Lightning AI

Step 2: Create new Studio

Step 3: Select “Code”

Hands-On Testing Step by Step

Get the Code

Step 1: Open the terminal and execute the following command:

​

wget https://github.com/hyperledger-labs/aifaq/archive/refs/heads/main.zip

​

Step 2: Unzip and remove main.py and main.zip

Step 3: Move into “aifaq-main/src/core” folder

Hands-On Testing Step by Step

Get the Document Source

Step 1: Get the Hyperledger Fabric ReadTheDocs:

​

wget -r -A.html -P rtdocs https://hyperledger-fabric.readthedocs.io/en/release-2.5/

​

Step 2: Stop (CTRL + C) when it is downloading other releases:

​

​

​

Step 3: move into “release-2.5” folder and compress the content:

tar -czvf rtdocs.tar.gz .

Step 4: Move the file in “rtdocs” folder

​

​

Hands-On Testing Step by Step

Switch to GPU

Hands-On Testing Step by Step

Create the Knowledge Base

Step 1: Install dependencies:

pip install -r requirements.txt

​

Step 2: Execute “ingest.py” script:

​

​

​

Hands-On Testing Step by Step

Test the System

Step 1: Run API:

​

Step 2: Execute “api.py” script:

​

​

​

Step 3: Check it:

​

Hands-On Testing Step by Step

Ask Questions

Step 1: Send a question via POST request:

curl --header "Content-Type: application/json" --request POST --data '{"text": "How to install Hyperledger fabric?"}' http://127.0.0.1:8080/query

​

​

​

​

Presentation Links