Vibeleaderboard
Index / tool
Visit github.com
Category
AI Tools
Pricing
Open Source
Type
TOOL
Builder
karpathy
Added
Apr 26, 2026

About

A minimal, fast implementation for training and fine-tuning medium-sized GPT models from scratch. Features clean, readable code with just ~300 lines each for the training loop and model definition, making it easy to customize and experiment with.

Why it made the leaderboard

Karpathy's minimal GPT training code — the training loop and model definition are ~300 lines each, yet it reproduces GPT-2 (124M) on OpenWebText in about four days on an 8xA100 node. Deprecated in favor of nanochat, but still the cleanest first read on how GPT training actually works.

Tags

gpttransformertrainingmachine-learningpytorchlanguage-modeldeep-learning

Featured in Intel

Comments (0)

No comments yet

Indexed by a proprietary survey. Corrections welcome.