nanogpt

Run GPT model on the browser with WebGPU. An implementation of GPT inference in less than ~1500 lines of vanilla Javascript.

This repository contains the collection of explorative notebooks pure in python and in the language that we, humans can read. Have tried to compile all lectures from the Andrej Karpathy's 💎 playlist on Neural Networks - which we will end up with building GPT.

Symbolic Music NLP Artificial Intelligence Toolkit

LLM frontend for creative writing and roleplay

A GPT built and trained in 243 lines of pure, dependency-free Python - no PyTorch, no NumPy. Based on Andrej Karpathy's microgpt, it hand-rolls autograd, attention and Adam to show how little a language model really needs.

LiteGPT: A 124M Small Language Model (SLM) pre-trained on FineWeb and fine-tuned on Alpaca.

A compact, readable PyTorch reference implementation of the DeepSeek-V4 architecture (Flash and Pro), in the spirit of nanoGPT. ~4,300 LOC.

burn inference and training of dragon models 🔥🐉

Interact with NanoGPT's API for pay-per-prompt interaction with AI models

Repository nanoGPT from @karpathy accelerated with JAX/Flax! The simplest, fastest repository for training/finetuning medium-sized GPTs.

Code repository for the paper "Traveling Words: A Geometric Interpretation of Transformers"

免费 AI / 深度学习学习资料合集 + 学习笔记与代码 | A curated list of free AI & deep learning resources (ML, DL, RL, GNN, LLM, Transformer, Diffusion), plus study notes and code (nanoGPT, minimind, VAE).

Model created by following the "Let's build GPT: from scratch, in code, spelled out." lecture by Andrej Karpathy, in Org Mode.

Send a prompt to multiple LLMs / text models / image models simultaneously

Production-Level Large Language Model Training & Deployment Framework

This repo represents my Nano-GPT speedrun playground, which started coding along Let's reproduce GPT-2 (124M), then moved into further improvements.

A secure multi-provider, multimodal Discord AI room harness for NanoGPT, OpenAI, Anthropic, xAI, and Gemini.

The simplest masked-diffusion language model you can actually train, debug, and learn from — ~1100 lines of plain PyTorch, char-level, with an honest head-to-head against a matched autoregressive baseline. Watch text crystallize from noise.

A 134M-parameter GPT written from scratch in PyTorch, no nn.Transformer and no HuggingFace, pretrained on a 7B tokens corpus of FineWeb-Edu, where it consumed 1.2B tokens on one free Kaggle P100. Held-out perplexity 36.12 against a Chinchilla prediction of 38.70.

Visual AI workflow editor in your browser — wire text, image, video, and audio models into node graphs and shareable apps. No backend, no analytics, bring your own key.