Practical AI Projects

Collection of practical AI applications and implementations using various machine learning, deep learning, and natural language processing techniques.

Overview

This repository contains a diverse set of practical AI projects that demonstrate the application of AI in various domains. The projects cover a wide range of techniques including:

• Large language model (LLM) training fine-tuning
• Retrieval Augmented Generation (RAG)
• Multi-agent AI systems
• Computer vision applications
• Natural language processing
• Sentiment analysis
• Recommendation systems
• Clustering and dimensionality reduction

Project Structure

Language Model Training and Fine-tuning

• training-large-language-model-for-esperanto.ipynb: Training a RoBERTa-like large language model (LLM) from scratch for Esperanto using custom tokenization and masked language modeling.
• language_model_finetuning_qlora_llama2.ipynb: Fine-tuning Llama 2 using QLoRA technique.
• finetuning_img_classifier_visual_transformer_lora.ipynb: Fine-tuning a visual transformer model using LoRA.

Retrieval Augmented Generation (RAG)

• retrieval_augmented_generation_agent_llm_llamaindex_gpt-4o.py: Implements a RAG agent using LlamaIndex and GPT-4o.
• graph_retrieval_augmented_generation_graph_rag_gpt-4o.py: Graph-based RAG implementation using GPT-4o.
• retrieval_augmented_generation_query_engine_llamaindex_rag_llm.py: Query engine for RAG using LlamaIndex.

Multi-Agent Systems

Located in the multi-agent_coding_stock-analysis_customer-onboarding_chess_writing_conversation_autogen directory:

• Multi-Agent_Coding_and_Financial_Analysis_AutoGen.ipynb: Demonstrates collaborative AI agents for coding and financial analysis.
• Multi-Agent_Conversation_and_Stand-up_Comedy_AutoGen.ipynb: AI agents generating conversational content and comedy.
• Multi-Agent_Planning_and_Stock_Report_Generation_AutoGen.ipynb: Agents that plan and generate stock reports.
• Multi-Agent_Reflection_and_Blogpost_Writing_AutoGen.ipynb: Collaborative writing and reflection through AI agents.
• Multi-Agent_Sequential_Chats_and_Customer_Onboarding_AutoGen.ipynb: Customer onboarding flows using multiple agents.
• Multi-Agent_Tool_Use_and_Conversational_Chess_AutoGen.ipynb: Tool use and chess game analysis by AI agents.
• Multi-Agent_Tool_Use_Fake_Nvidia_Stocks.py: Demonstration of tool use for stock analysis.

Computer Vision

• natural-scene-classification_cnns.ipynb: Classification of natural scenes using CNNs.
• image-classifier_resnet18.ipynb: Image classification using ResNet18.
• image-classifier_resnet34.py: Image classification using ResNet34.
• segmentation_resnet34.py: Image segmentation using ResNet34.
• noise_removal_autoencoder.py: Noise removal using autoencoders.
• image_question_answering_openai.py: Visual question answering using OpenAI's GPT-4o.
• text-to-image_clipdrop.py: Text-to-image generation using ClipDrop.
• resnet_transfer_learning_pneumonia.py: Transfer learning with ResNet for pneumonia detection.

Natural Language Processing

• NLP_classification_search_text-edit_sentiment_analysis_Bert_BoW_SVM_embeddings_skip-gram_regex_POS.ipynb: Comprehensive NLP techniques.
• sentiment-analysis_vader_roberta.ipynb: Sentiment analysis using VADER and RoBERTa.
• sentiment_analysis_AWD-LSTM_enocder.py: Sentiment analysis using AWD-LSTM encoder.
• question_answering_openai.py: Question answering using OpenAI models.
• text-to-speech_elevenlabs.py: Text-to-speech conversion using ElevenLabs.

Other

• clustering_kmeans.ipynb: K-means clustering implementation.
• dimensionality_reduction_pca.ipynb: Principal Component Analysis (PCA) for dimensionality reduction.
• tabular_prediction_nn.py: Neural network for tabular data prediction.
• digit-classifier_nn.py: Neural network for digit classification.
• recommender_system_movies_nn.py: Movie recommendation system using neural networks.
• 2d_flow_matching_checkerboards.ipynb: Implementation of 2D flow matching for generating checkerboard patterns.

Getting Started

Prerequisites

• Python 3.8+
• Required packages depending on the specific project (see individual files)
• API keys for various services (OpenAI, ElevenLabs, etc.)

Installation

1. Clone this repository:

   git clone https://github.com/yourusername/practical-ai-projects.git
   cd practical-ai-projects

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies for the specific project you want to run.

Usage

Each file or notebook contains example usage. For Python scripts, you can run them directly:

python retrieval_augmented_generation_agent_llm_llamaindex_gpt-4o.py

For Jupyter notebooks, open them in Jupyter Lab or Notebook:

jupyter lab

API Key Setup

Many of these projects require API keys. Create a .env file in the root directory with the following content:

OPENAI_API_KEY=your_openai_api_key
ELEVENLABS_API_KEY=your_elevenlabs_api_key
# Add any other API keys required

License

[MIT License]

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Acknowledgments

• OpenAI for GPT models
• LlamaIndex for RAG implementations
• AutoGen for multi-agent systems
• Various other open-source libraries and frameworks used in the projects