This repository provides a curated AI crash course designed for busy builders and founders to quickly grasp the frontier of AI research within two weeks. It offers a structured learning path, prioritizing key papers, surveys, and resources to bridge the knowledge gap for those transitioning into AI development.

How It Works

The course follows a progressive learning structure, starting with foundational neural network concepts and progressing to Large Language Models (LLMs). It emphasizes understanding through survey papers and then deep-diving into seminal research papers, categorized by sub-fields like foundational modeling, planning/reasoning, applications, and benchmarks. This approach allows users to build a solid understanding of core AI advancements and their practical implications.

Quick Start & Requirements

This is a curated list of resources, not a software package. No installation is required. Access to the internet is necessary to view the linked papers, videos, and websites.

Highlighted Details

• Covers foundational papers like "Attention Is All You Need" (Transformers) and "Scaling Laws for Neural Language Models" (GPT-3).
• Includes recent advancements in LLM reasoning (Chain of Thought, Tree of Thoughts) and agentic systems (ReAct, SWE-Agent).
• Features practical applications and overviews of models like Llama 3, Gemini 1.5, and Deepseek V3.
• Provides links to video lectures from prominent researchers like Andrej Karpathy and Yannic Kilcher.

Maintenance & Community

The project is maintained by Henry Shi. Further community interaction or roadmap details are not explicitly mentioned in the README.

Licensing & Compatibility

The repository itself is likely under a permissive license given its nature as a curated list. However, the linked research papers and resources are subject to their original licenses and copyrights.

Limitations & Caveats

This resource is a guide to understanding research and does not provide code or tools for direct implementation. The rapidly evolving nature of AI means some information may become dated quickly, requiring users to supplement with the latest research.