MOLI

Results 3 issues of MOLI

作者您好,我在将G-ghostNet改进到自己的框架中时,原论文中描述的Mix操作是先进行全局平均池化后再经过一个全连接层,但是代码中是图中所示(希望我没找错)![mix操作](https://github.com/huawei-noah/Efficient-AI-Backbones/assets/102312535/8a4b27ba-d398-4cb1-a18d-5e4c8f8f15a7),请您解释一下此处,万分感谢!

请问g-ghostnet有没有pytorch的预训练权重呢?

For the STACP code I ran, the average precision of all users(Precision@10) calculated using the Gowalla dataset is only 0.021179. Do I need to change the source code to reach...