nanochat

Autonomous AI research swarm — runs ~100 overnight LLM training experiments unattended. Fork of karpathy/autoresearch with Digital Cognitive Labor routing.

Train your own ChatGPT on Apple Silicon — MLX port of nanochat

Train Llama 3 models from scratch. Any scale, any personality. By Arianna Method.

A minimal, hackable Vision-Language Model built on Karpathy’s nanochat — add image understanding and multimodal chat for under $200 in compute.

A user's facts become a few rows in a content-addressed memory table — not a rewrite of the model. Paper, code, data, and interactive site for "User as Engram" (arXiv:2606.19172).

The best ChatGPT-style model that $100 of TPU time can buy.

nanochat's inference engine re-vibed in C++ with GGML.

Ascend NPU fork of nanochat for LLM training with torch_npu/HCCL (experimental)

A legible and hackable port of Nanochat in JAX/Flax

Erk — eCloud Tech. Türkçe LLM ailesi. nanosohbet: Erk modellerini sıfırdan eğiten tam açık reçete (Türkçe nanochat).

I see now that the path I choose through the maze makes me what I am.

The official implementation of Ringmaster LMO, an asynchronous distributed optimizer for neural network training under heterogeneous compute environments.

A 1.68B chat model trained from scratch (pretraining + SFT + GRPO) on the nanochat recipe. Base CORE 0.30 vs GPT-2 0.26, with a logit-level analysis of the RL checkpoint's regressions.

Production-honest small language model training factory: data import, pretraining, SFT, eval gates, contamination checks, and GPU runbooks.

llama.cpp fork adding nanochat architecture support, so Karpathy's nanochat checkpoints run as GGUF. See NANOCHAT.md

The best GPT that $100-$125 worth of pre-training and finetuning can buy

I built this repo to prove to my granny that I can implement GPT.

The best ChatGPT that $100 can buy ported to an Nvidia RTX 5090 which is NOT $100

Autonomous ML research on a single DGX Spark: three agent modes spanning model capability (hypothesis generator to Karpathy's original design), custom GGUF support for frontier-scale quants like DeepSeek V4 Flash, a 4-level model/harness benchmark suite, and game-theoretic multi-branch orchestration

Implementing Karpathy's nanochat in rust.