tongjuntx

Results 4 comments of tongjuntx

according to my understanding: 1.the purpose of model without TSA model is to initialize weights of the complete model with TSA 2.stage2 is just for fine-tuning,both moderate model and lager...

> 谢谢大佬的回复,确实是很厉害的工作和研究。 > > 我们正在尝试先把可变卷积换成正常的卷积,然后训练得到的初始model,然后用这个模型训练网络。 > 接着冻结部分模型块再开始训练。 have you succeed?how about the effect?

the same question,some area can not be completed,it's so strange,increase patch_size?