tangchen2
tangchen2
> > 因此loss function (1-y) * (1-log（p)) 消除掉了 > > 这句话不是很懂，困扰了我好久了，还希望你能在详细解释一下~ > > 蛮烦了~ > > @OceanWong1991 > > 这个一定要结合后面看. > 在loss=-tf.log(label_prob+1e-10)时候的 > > 1. Label_prob包含 “neg label”取neg的概率 > 2....
感谢回复，试了一下bottleneck内部的conv - bn merge，提速效果收益很小
> how to extract the relation and entity for a sentence？ it can just predict the relation type such as message-top and so on, it can't predict the entity in...
我没有用mobilenet的backbone而是大模型在预训练的基础上训练的，数据集2007+2012，目前大概20个epoch左右，mAP在62.4%左右，训练的loss在13左右，现在开始loss下降的速度比较缓慢，整体情况和@Imagery007 比较类似
> 你好，我也遇到这个问题，请教一下。我程序里是可以跳到“revertMNNModel.hpp”这个头文件的，那是否代表能找到这个头文件？ 你好，请问你解决这个问题了么，我程序里包含了“revertMNNModel.hpp”，但仍然无法链接到Revertor
> @tangchen2 Have you used the pretrained weights for FaceNet, as they mention in the paper? Otherwise, the results are likely quite a bit worse if you are training only...
> Another thing worth checking is that the order of the color channels that you input to FaceNet (RGB versus BGR). Well, i really didnt concern about the color channel,...
> Last (obvious) thing to check is that you are only considering what they refer to as "strong" triplets, i.e., the ones where at least 66% of the raters agree...
> Last (obvious) thing to check is that you are only considering what they refer to as "strong" triplets, i.e., the ones where at least 66% of the raters agree...
> Because of the loss of crawler data, I cannot have all the triple pairs. How can I get the original data. l appreciate that you can share the data...