AI and Machine Learning Projects

This repository contains a collection of five projects showcasing different AI and machine learning technologies and frameworks. Each project focuses on a unique aspect of AI, such as natural language processing, text generation, or audio transcription.

---

Projects Overview

1. Llama_cpp: Basic Chat Completion

• Description: Utilizes the llama_cpp library to load a pre-trained Llama model and perform chat-based text generation.
• Key Features:
  • Supports GPU acceleration and customizable parameters like context length and random seed.
• Script: Demonstrates creating a conversational assistant for answering questions.

Learn More

---

2. Llama_cpp with Parallel Processing

• Description: Extends the first project by leveraging ray for parallel execution of multiple tasks using the Llama model.
• Key Features:
  • Enables parallel chat completion tasks for improved efficiency.
• Script: Handles multiple Llama chat completions simultaneously.

Learn More

---

3. Text Generation with Meta-Llama

• Description: Implements text generation using Hugging Face's transformers library with the Meta-Llama-3-8B model.
• Key Features:
  • Supports mixed-precision computation (bfloat16) and automatic device mapping.
• Script: Generates coherent responses to user prompts.

Learn More

---

4. Audio Transcription with Whisper

• Description: Transcribes audio into text using OpenAI's Whisper model.
• Key Features:
  • Supports multiple Whisper model types.
  • Converts PyTorch tensors to NumPy arrays for compatibility.
• Script: Includes a Transcription class for processing audio waveforms.

Learn More

---

Installation

Prerequisites

• Python 3.8 or higher
• Additional libraries required for each project (listed in individual READMEs).

General Setup

Clone the repository:

git clone <repository-url>
cd <repository-folder>

Install necessary libraries:

pip install -r requirements.txt

---

Usage

Each project has its own README file with detailed setup and execution instructions. Refer to the respective links provided above for specific guidance.

---

Notes

• Ensure proper hardware (e.g., GPUs) for models requiring high computational resources.
• Some projects may require additional configurations, such as downloading pre-trained models.

---

License

This repository is licensed under the MIT License. Please refer to the LICENSE file for details.