Technology Radar
Published : Apr 02, 2025
NOT ON THE CURRENT EDITION
This blip is not on the current edition of the Radar. If it was on one of the last few editions, it is likely that it is still relevant. If the blip is older, it might no longer be relevant and our assessment might be different today. Unfortunately, we simply don't have the bandwidth to continuously review blips from previous editions of the Radar.
Understand more
Apr 2025
Assess
torchtune 是一个专为 PyTorch 设计的库，用于编写、后训练以及实验性探索大语言模型。它支持单 GPU 和多 GPU 设置，并通过 FSDP2 实现分布式训练。该库提供基于 YAML 的 recipes（配方），用于微调、推理、评估以及量化感知训练等任务。每个配方都聚焦于特定功能，避免复杂的参数标志配置，注重代码清晰性而非过度抽象化。此外，torchtune 包含一个强大的 CLI，可用于高效地下载模型、管理配方和运行实验。
