Llama-Custom-X - Chatbot Assistant 🤖

This repository contains a Streamlit-based web application named Llama-Custom-X, a Cyber Assistant designed to interact with CSV datasets and provide data-driven insights through a chatbot interface. The app leverages machine learning models to facilitate conversational queries and offers robust data visualization features.

Features

• Chatbot Interface: Conversational interaction with the first dataset (CQS.csv). Ask questions about the data, and the AI will respond with insightful answers.
• Data Preview and Visualization: Visualize key aspects of the datasets (CQS.csv and LOG.csv) using count plots, time series, ring charts, and more.
• Customizable and Expandable: Easy to add more visualizations or modify existing ones based on your dataset requirements.

Installation

1. Clone the repository:

   git clone https://github.com/Ayushverma135/Llama-Custom-X.git
   cd Llama-Custom-X

2. Install dependencies: Make sure you have Python 3.8+ installed. Then, install the required packages:

   pip install -r requirements.txt

3. Download the model: Ensure that the llama-2-7b-chat.Q4_0.gguf model is available in your system. You can download it from GPT4All or Hugging Face.

Usage

1. Run the application:

   streamlit run app.py

2. Interact with the Cyber Assistant:

   • Use the chatbot to query your data in the CQS.csv file. The assistant provides intelligent answers based on the CSV content.
   • Explore visualizations and statistics for both CQS.csv and LOG.csv.

Dependencies

• streamlit - Web framework for creating the interactive app.
• pandas - Data manipulation and analysis library.
• matplotlib & seaborn - For data visualization.
• langchain - Framework for building language model-powered applications.
• HuggingFaceEmbeddings - To embed text for similarity search.
• FAISS - Facebook AI Similarity Search, used for fast similarity search.

Installing llama-2-7b-chat.Q4_0.gguf Model from GPT4All and Hugging Face

From GPT4All:

1. Install GPT4All:

   • If you haven't already, you can install GPT4All with the following command:

     pip install gpt4all

2. Download the Model:

   • Visit the GPT4All website to download the llama-2-7b-chat.Q4_0.gguf model. Ensure you download the .gguf version.
   • Alternatively, you can use a direct download link (if provided by GPT4All) to download the model.

3. Load the Model:

   • Once the model is downloaded, you can load it using GPT4All in Python:

     from gpt4all import GPT4All
     
     model_path = "path/to/llama-2-7b-chat.Q4_0.gguf"
     gpt = GPT4All(model_path)

From Hugging Face:

1. Install Hugging Face Transformers:

   • Install the transformers library if you don't have it:

     pip install transformers

2. Install Hugging Face Hub (Optional):

   • If you want to interact with models directly from the Hugging Face Hub:

     pip install huggingface_hub

3. Download the Model:

   • Use the Hugging Face model page to download llama-2-7b-chat.Q4_0.gguf. You may need to specify the correct variant if it’s available.
   • Alternatively, you can clone the model repository:

     git lfs install
     git clone https://huggingface.co/your-username/llama-2-7b-chat

4. Load the Model:

   • Load the model using the transformers library:

     from transformers import AutoModelForCausalLM, AutoTokenizer
     
     model_name = "path/to/llama-2-7b-chat"
     tokenizer = AutoTokenizer.from_pretrained(model_name)
     model = AutoModelForCausalLM.from_pretrained(model_name)

Acknowledgements

• Streamlit for providing an easy-to-use framework for building interactive web apps.
• LangChain for making it easier to build language model applications.
• Hugging Face for their comprehensive library of machine learning models and embeddings.
• FAISS for enabling efficient similarity searches over large datasets.

Notes:

• The specific path or way to load the model might vary depending on updates from GPT4All or Hugging Face, so always refer to the latest documentation.
• Ensure your environment has sufficient resources to handle large models like llama-2-7b-chat.