Res-Tuning 是一种灵活高效的微调tuner。我们把tuner的设计从模型网络结构中解耦出来以便灵活地组合, 并进一步扩展实现了一种新的节省内存的旁路tuner,大大减少了显存消耗和多任务推理成本。
目前Res-Tuning在SWIFT中以可插拔的tuner算法组件提供,开发者可以直接使用它。
- Res-Adapter
- Res-Tuning-Bypass
- Res-Prefix
- Res-Prompt
- 可以使用我们提供的 可视化例子.
from swift import ResTuningConfig
config = ResTuningConfig(
dims=768,
root_modules=r'.*blocks.0$',
stem_modules=r'.*blocks\.\d+$',
target_modules=r'norm',
tuner_cfg='res_adapter'
)
- dims: The dimensions of the hidden states.
- root_modules: The root module to be replaced.
- stem_modules: The stem modules to be replaced.
- target_modules: The target module to be replaced.
- tuner_cfg: The configuration of the tuning module.
from swift import Swift
import timm, torch
model = timm.create_model("vit_base_patch16_224", pretrained=False, num_classes=100)
model_tune = Swift.prepare_model(model, config)
print(model_tune.get_trainable_parameters())
print(model(torch.ones(1, 3, 224, 224)).shape)
@inproceedings{jiang2023restuning,
title={Res-Tuning: A Flexible and Efficient Tuning Paradigm via Unbinding Tuner from Backbone},
author={Jiang, Zeyinzi and Mao, Chaojie and Huang, Ziyuan and Ma, Ao and Lv, Yiliang and Shen, Yujun and Zhao, Deli and Zhou, Jingren},
booktitle={Advances in Neural Information Processing Systems},
year={2023}
}