borgwang/tinynn

Issues

• 激活函数问题

  #32 opened 2 years ago by Mylszd
  4

• 可以添加transformer ，seq2seq 和 attention 这些吗？

  #29 opened 3 years ago by zuowanbushiwo
  3

• About Huber Loss

  #26 opened 3 years ago by Zhangang1999
  2

• SOFTPLUS

  #25 opened 3 years ago by Zhangang1999
  1

• Step should average the gradients by batch size.

  #12 opened 5 years ago by w32zhong
  8

• One suggestion on file header

  #5 opened 5 years ago by w32zhong
  3

• Avoid download dataset programmatically.

  #1 opened 5 years ago by w32zhong
  3

• The MNIST example has surprising test accuracy.

  #2 opened 5 years ago by w32zhong
  2