M1ngzzZ

The discriminator loss function your used in paper us is" min(0,1-D_k(x))+ min(0,1+D_k(G(s,z)))" ,but u said in your paper it is hinge loss ,and u used the Relu method in your...

您好，我复现了下您的代码，一直没有达到理想的想过想请假您一些问题？

4

我最近复现了一下您的代码，核心代码都一样，都是从您DCASE2020的代码copy的，然后网络结构、参数跟您设置的一致、Encoder的预训练参数也是down的您的模型，但是实验结果一直很差，Spiderscore可能才0.18左右，想请教您一下原因，真的调了好久了

Unable to reproduce the results of the paper

2

Hello. I tried using the demo code of Codi (https://github.com/microsoft/i-Code/tree/main/i-Code-V3) to reproduce results on the AudioCaps dataset. However, I was unable to achieve the results reported in the paper for...