tianchen
tianchen
@jindongwang 请问作者您尝试过复现AdaBN的结果吗?AdaBN的论文和后来诸如VisDA等比赛中都常用到这种方法,但是经过我自己复现之后发现效果并不是很好)☹ 如果您有相关研究经历,可否与我解惑,感激不尽🙏
I have the same problem here..don't seem to understand how to transform the current data to the smaller one.n@may0324
@liu666666 Hey Bro, I found one Quantization tool which could compress the .caffemodel into the .npz(stored in data, Codebook, labels) at https://github.com/yuanyuanli85/CaffeModelCompression, This solved my problem, probably could solve yours.