Welcome to the Intro2AGI repository! This repo is organized into two main projects, each catering to different audiences and objectives:
- ENGR-102: Visual Reasoning & Wranglin’ GPTs
- ECE-401: Neural Networks, Transformers, and Multimodal Systems
Below, you'll find an overview of each folder, including its purpose, key contents, and learning objectives.
Overview:
This project introduces beginning students to the fascinating world of Large Language Models (LLMs) and Artificial Intelligence (AI). No prior coding or AI experience is required—just curiosity and a willingness to explore! The course focuses on developing systems thinking, mental modeling, and communication skills while leveraging AI tools like ChatGPT for creative problem-solving.
Key Files:
Presentation.pdf: A fun and engaging introduction to AI concepts, focusing on visual reasoning and using LLMs to solve problems systematically.Assignment_Memo.pdf: Step-by-step guidance for students to dive into tasks from the ARC AGI Challenge, including learning objectives and hands-on exercises.
Learning Objectives:
- Systems Thinking: Learn to identify patterns, break down problems, and solve challenges systematically.
- Mental Modeling: Reflect on problem-solving approaches and refine strategies for effective reasoning.
- Communication: Master clear explanations of complex ideas, translating them into actionable steps or "algorithms."
- AI Cowboycraft: Wrangle ChatGPT or similar LLMs to transform written steps into Python code.
- Scalable Thinking: Design solutions that generalize across tasks rather than relying on hard-coded answers.
For more, explore the ARC-AGI Challenge via ARC Puzzle Tasks.
Overview:
This advanced project is geared toward personal exploration and learning. It dives into neural network fundamentals, the transformer architecture, and multimodal systems. Students and researchers will tackle visual reasoning tasks from the ARC dataset and prototype systems architectures aimed at Artificial General Intelligence (AGI).
Key Files:
Syllabus.pdf: Detailed course outline covering the ARC Challenge, neural network fundamentals, and multimodal system integration.
Learning Objectives:
- Neural Network Fundamentals: Complete Andrej Karpathy’s "Neural Networks: Zero to Hero" lecture series to build a strong foundation.
- Visual Reasoning and Intuition: Design systems for visual reasoning, integrating multimodal inputs for generalizable problem-solving.
- Systems Thinking and Architecture Development: Prototype scalable architectures with transformers and multimodal neural networks.
- ARC Challenge Submission: Develop a practical solution to the ARC AGI Challenge, showcasing innovative system design and reasoning.
Evaluation Criteria:
- ARC Submission (55%): A comprehensive solution, including a detailed write-up and final code.
- GitHub Updates (30%): Weekly progress in the form of documented code and reflection reports.
- Synthesis Reports (15%): Analytical insights connecting coursework to AGI research goals.
Feel free to explore both projects and contribute to the journey of understanding and advancing AI! 🚀