【文件属性】：
文件名称：MADGRAD-Segmentation-Pytorch-Performance:比较MADGRAD与Adam和SGD在分割模型训练中的表现
文件大小：191KB
文件格式：ZIP
更新时间：2021-04-05 22:26:49
Python
MADGRAD分段分割火炬性能
Compare the performance of MADGRAD against Adam and SGD for segmentation model training.
交叉熵损失
像素精度
VOC2012细分培训
- SDG
~/anaconda3/envs/pytorch17_102/bin/python train.py SGD
- Adam
~/anaconda3/envs/pytorch17_102/bin/python train.py Adam
- MADGRAD
~/anaconda3/envs/pytorch17_102/bin/python train.py MADGRAD
参考
- https://github.com/jfzhang95/pytorch-deeplab-xception.git (
【文件预览】：
MADGRAD-Segmentation-Pytorch-Performance-main
----.gitignore(2KB)
----README.MD(912B)
----customUNet.py(11KB)
----read_tensorboard_out.py(2KB)
----config.py(285B)
----metrics.py(2KB)
----dataloaders()
--------datasets()
--------__init__.py(2KB)
--------custom_transforms.py(5KB)
--------utils.py(3KB)
----custom_utils.py(109B)
----figures()
--------crossentropyloss_MADGRAD.png(44KB)
--------pixelaccuracy_MADGRAD.png(24KB)
--------crossentropyloss_ADAM.png(38KB)
--------pixelaccuracy_ADAM.png(23KB)
--------pixelaccuracy_SGD.png(21KB)
--------crossentropyloss_SGD.png(30KB)
----pascal.py(5KB)
----train.py(11KB)