How stevengonsalvez/llm-dojo works
Main Components Organization
┌─ Part 1: Understanding LLMs ─┐ ┌─ Part 2: Advanced Prompt Engineering ─┐
• Models & Tokens │ │ - Prompt Engineering Components │
• Classification Systems │ │ - Evaluation Systems │
• Chain of Thought │ │ - N-shot & RAG Components │
• ┘ └────────────────────────────────────────┘
┌─ Part 3: Frameworks/Models ─┐ ┌─ Part 4: Beyond Text ─────────────────┐
• Langchain Integration │ │ - Diffusion Models │
• Model Implementations │ │ - Training Pipeline │
• Platform Services │ │ - Custom Data Processing │
• ┘ └────────────────────────────────────────┘
Technical Components
a) Infrastructure Layer:
• Jupyter Notebook environments (both local and Google Colab)
• Environment management system (conda/virtualenv)
• Authentication & API key management
b) Core Components:
• GPT API Integration (from gpt-api directory)
• Langchain Framework (from langchain directory)
• Training Components (from training directory)
• Service Application (from apps/service-app)
Integration Points
• Data flow between notebooks and external APIs
• Environment variable management system
• Integration with Google Drive for Colab setup
• Connection to various LLM services (OpenAI, PALM2, etc.)
Specific Implementation Details
┌─ Environment Management ─┐
• Local .env │
• Colab Drive Mount │
• API Key Management │
• ┘
┌─ Notebook Environments ─┐
• Local Jupyter │
• Google Colab │
• VSCode Integration │
• ┘
Key Relationships to Highlight
• Data flow for training components
• Security boundaries around API key management
• Integration points with external frameworks