Educational GPT implementation in ~300 lines. Reproduces GPT-2 (124M) on OpenWebText. Clean, hackable code for learning transformers. By Andrej Karpathy. Perfect for understanding GPT architecture from scratch. Train on Shakespeare (CPU) or OpenWebText (multi-GPU).
7.0
Rating
0
Installs
AI & LLM
Category
Excellent educational skill for GPT training with comprehensive workflows, clear examples, and proper structure. The SKILL.md provides complete command sequences for multiple use cases (Shakespeare training, GPT-2 reproduction, fine-tuning, custom datasets) with realistic outputs and troubleshooting. Documentation is well-organized with references separated appropriately. The description accurately reflects capabilities and a CLI agent could invoke any workflow based solely on the main documentation. Novelty score is moderate because while the skill packages educational code nicely, the underlying simplicity (intentionally ~300 lines) means a capable agent could potentially construct similar solutions, though this skill does save significant setup and experimentation time. Minor deductions only for the novelty dimension - all other aspects are exemplary for an educational implementation skill.
zechenzhangAGI
Skill Author
Loading SKILL.md…
