文件名称:revnet-public:“可逆残差网络”的代码
文件大小:493KB
文件格式:ZIP
更新时间:2024-05-31 23:30:31
Python
公开网 纸质代码可逆残留网络:不存储激活的反向传播。 [ ] 安装 首先在setup.sh自定义路径(数据文件夹,模型保存文件夹等)。 git clone git://github.com/renmengye/revnet-public.git cd revnet-public # Change paths in setup.sh # It also provides options to download CIFAR and ImageNet data. (ImageNet # experiments require dataset in tfrecord format). ./setup.sh CIFAR-10 / 100 ./run_cifar_train.py --dataset [DATASET] --model [MODEL] DATASET可用值为cifar-10和cif
【文件预览】:
revnet-public-master
----.gitignore(1KB)
----run_all_unittests.sh(86B)
----run_imagenet_eval.py(5KB)
----LICENSE(1KB)
----run_cifar_train.py(9KB)
----tools()
--------download_and_preprocess_imagenet.sh(4KB)
--------imagenet_lsvrc_2015_synsets.txt(10KB)
--------process_bounding_boxes.py(9KB)
--------imagenet_2012_validation_synset_labels.txt(488KB)
--------preprocess_imagenet_validation_data.py(3KB)
--------download_imagenet.sh(3KB)
--------download_cifar.sh(636B)
--------build_imagenet_data.py(26KB)
--------imagenet_metadata.txt(724KB)
----README.md(2KB)
----setup.sh(1KB)
----resnet()
--------data_tfrecord()
--------data()
--------__init__.py(0B)
--------models()
--------utils()
--------configs()
----run_imagenet_train.py(9KB)