复现该项目的代码失败:aimagelab/meshed-memory-transformer
复现时参考该博主的文章:meshed-memory transformer代码实现(绝对详细)(然而还是未成功!)
复现时遇到的问题以及解决方案(供参考,有人复现成功啦,评论告诉一下哈,想参考参考呢)
参考链接:
1、安装pycocotools失败解决方法:
在windows环境下安装pycocotools(这种方法下载的文件需要放在C:\Users\deep\anaconda3\Lib\site-packages下)
安装
2、pycocotools安装报错解决方案:pycocotools安装报错解决
3、报错ModuleNotFoundError:No module named‘past’:ModuleNotFoundError:No module named‘past’解决方法
安装成功:
(m2release) D:\***\meshed-memory-transformer-master>pip install future==0.17.1
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting future==0.17.1
Downloading https://pypi.tuna.tsinghua.edu.cn/packages/90/52/e20466b85000a181e1e144fd8305caf2cf475e2f9674e797b222f8105f5f/future-0.17.1.tar.gz (829kB)
|████████████████████████████████| 829kB 6.4MB/s
Building wheels for collected packages: future
Building wheel for future (setup.py) ... done
Created wheel for future: filename=future-0.17.1-cp36-none-any.whl size=488736 sha256=9cd9cf45f8318832d9ea04007b5ec5bbbd238bca95a7b7a49833bc9896de879a
Stored in directory: C:\Users\deep\AppData\Local\pip\Cache\wheels\3f\66\fe\9c4fd5c707a9f26993ba157f0752d84d5c7e26aedbefb84f76
Successfully built future
Installing collected packages: future
Successfully installed future-0.17.1
4、torch的离线下载网站:网站
5、明明有h5py包但是运行程序时还是找不到h5py库:h5py库安装问题解决