Learning rate Warmup

Creator
Creator
Seonglae ChoSeonglae Cho
Created
Created
2024 Mar 1 14:13
Editor
Edited
Edited
2024 Mar 3 4:19
Refs
Refs
initialization
초기에는 작은 learning rate 사용하다 안정되면 높이는 방식
  • warmup_steps Number of steps used for a linear warmup from 0 to learning_rate
  • warmup_ratio Ratio of total training steps used for a linear warmup from 0 to learning_rate.
 
 
 
 
 
 

Recommendations