lostmsu/TurboGPT

Train a tiny GPT in under a minute (CUDA only)

Language modelsC++

2.0 Steady Magnitude out of 10 — how fast interest is growing, not a quality score. Star growth looks organic. Data as of September 29, 2026, 22:17 UTC.

Open on Seismograph Open on GitHub

About the project

Trains a tiny byte-level GPT from scratch in CUDA C++ in under a minute; suited for experimenting with language model training on a local GPU.

Useful for

Train a mini-GPT on a custom text file and get a checkpoint for further fine-tuning
Resume interrupted training from a saved checkpoint using --load
Check model quality via BPB metric on the hn1g dataset

README summarized by DeepSeek V4.1 Flash. Details may be inaccurate.

Why it’s trending

10 stars today.
The repository is 6 days old and already has 9 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
Hacker News: “Show HN: TurboGPT: train 22KiB transformer in 13s” — 19 points, 2 h ago.

Stars per day

Bars are daily stars, the line is the usual pace. Red marks spike days.

Numbers

Total stars: 9
Today: 10
Forks: 0
Issues and pull requests: 0
Watchers: 0
Language: C++
License: MIT
Latest release: v0.0.1 · September 29, 2026
Created: September 23, 2026
Last push: September 29, 2026

Star trust

Growth looks organic: forks and discussion are in line with active projects, and stars arrive unevenly, the way people give them.

These are heuristics, not a verdict: we judge by the repository’s behavior, not by a list of stargazers.

Hacker News discussions

Show HN: TurboGPT: train 22KiB transformer in 13s19 points, September 29, 2026

Spotted in

September 29, 2026Spotted on Hacker News

README badge

Paste it into your README — the badge shows the current magnitude and links to this page.

[![Seismograph](https://gitnova.dev/badge/lostmsu/TurboGPT.svg?lang=en)](https://gitnova.dev/en/r/lostmsu/TurboGPT)

Similar by description

1.5
volotat/mini-AGI
A byte-level continual-learning language model that assembles its own architecture, trains from scratch on a single 8 GB VRAM GPU, and keeps learning from everything it reads. An experiment showing continual learning…

CoolingLanguage modelsPython+7 stars today
3.5
microsoft/SkillOpt
A text-space optimizer that trains reusable natural-language skills for frozen LLM agents via trajectory-driven edits, validation-gated updates, and a deployable best_skill.md artifact. Model weights stay untouched.

BreakoutAI agentsPython+52 stars today
2.5
jingyaogong/minimind
An open-source educational project for training a tiny 64M-parameter LLM (MiniMind) from scratch in PyTorch, covering the full pipeline: pretrain, SFT, LoRA, RLHF, RLAIF and Tool Use.

CoolingLanguage modelsPython+39 stars today
1.5
MrNeRF/LichtFeld-Studio
A native application to train, inspect, edit, and export 3D Gaussian Splatting scenes, with Python plugins and MCP-driven automation. Requires an NVIDIA GPU with CUDA 12.8+.

SteadyGenerative mediaC+++5 stars today