AI Agents Practice Repository

A collection of projects exploring different aspects of AI agent development.

Projects

1. 🤖 LLM Fundamentals Practice (000_llm_practice)

• Implementation of GPT from scratch using PyTorch
• Tokenization experiments with SentencePiece and tiktoken
• Dataset: Tiny Shakespeare

2. 🔍 Web Search Agent (001_web_search_agent)

• Terminal-based agent using LangGraph and DuckDuckGo Search
• Features tool usage and response formatting
• Implements a state machine for agent workflow

3. 📚 Agentic RAG System (002_agentic_rag)

• Retrieval Augmented Generation with LlamaIndex
• Custom retriever tool for guest information
• Ollama integration for local LLM operations
• For details, please check project section & Medium post!

Setup

1. Clone repository:

git clone https://github.com/your-username/agents-practice.git
cd agents-practice

2. Create conda environment:

conda env create -f 000_llm_practice/001_gpt_from_scratch/agentenv_backup
conda activate agentenv