TurboGPT
Train a tiny GPT in under a minute (CUDA only)
Train a tiny GPT in under a minute (CUDA only)
Train, fine-tune, and evaluate models — the infrastructure layer.
Key features
- 纯 CUDA C++ 实现
- 一分钟训完 tiny GPT
- 字节级建模
- TensorBoard 兼容日志
- 支持断点续训
- MIT 开源
Good for
- 学习 GPT 训练原理
- CUDA 编程实践
- 快速验证想法
- 教学演示
Stack & tags
CUDAC++
Quick start
.\build\turbogpt.exe --data hn1g.txt --log-to runs/ctx4From the project README on GitHub.
Similar open-source projects
tensorflow
★ 200,721An Open Source Machine Learning Framework for Everyone
transformers
★ 167,239🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.