yatengLG/Focal-Loss-Pytorch

Issues

• 请问如何设计正则化来防止focal loss带来的过拟合会比较好？可以帮忙给出一个样例参考吗？

  #20 opened 9 months ago by qianhuiliu
  0

• typos in your readme

  #19 opened 10 months ago by jizhang02
  1

• 当num_classes=2时（做二分类问题时），出现错误

  #18 opened a year ago by zxcvbnmkj
  0

• Expected object of backend CPU but got backend CUDA for argument #3 'index'

  #17 opened a year ago by wlsrick
  0

• 代码bug

  #5 opened a year ago by susoooon
  0

• 关于背景类

  #8 opened a year ago by aoko-sys
  5

• 关于α参数

  #4 opened a year ago by chenf99
  6

• 请问类别不平衡的解决思路是什么

  #3 opened a year ago by qwe3208620
  1

• CUDA error: an illegal memory access was encountered

  #6 opened a year ago by 18022443868
  2

• 数值不稳定问题

  #7 opened a year ago by yangwei-nlp
  1

• 当batch size是1的时候报错

  #15 opened a year ago by lx9cat
  1

• 这个报错该怎么解决？

  #10 opened a year ago by jiaoxiaosong
  1

• alpha bug

  #9 opened a year ago by suluo
  1

• 图像分割使用问题

  #11 opened a year ago by huweiv
  3

• 原始论文中提到，应该给样本数量较少的类别较小的权重

  #12 opened a year ago by wmathor
  12

• NER中num_class怎么设置？

  #13 opened a year ago by Smile-L-up
  0

• alpha如何设置？

  #2 opened a year ago by ChenXiao61
  19

• 这个第41行的self.alpha,应该不用更新吧？

  #14 opened 3 years ago by jjxyai
  1

• 老哥，第40行是不是应该用reshape啊？

  #1 opened 5 years ago by RichardHWD
  1