Vance0124/Token-level-Direct-Preference-Optimization

Issues

• Inquiry About KL Divergence in ICML 2024 Paper

  #8 opened 3 months ago by junkangwu
  2

• Some questions about the paper

  #7 opened 3 months ago by robin087
  1

• Can you train DPO directly? Using open-source base models.

  #6 opened 3 months ago by tcxia
  0

• How do you go about evaluating ALIGNMENT(accuracy) and DIVERSITY(entropy)? Is there a code available?

  #5 opened 5 months ago by wangxu0820
  0

• How to eval the models

  #4 opened 5 months ago by Yeeesir
  1

• How about the loss curve? especially when converge

  #3 opened 5 months ago by LuckerYi
  1

• Some questions about the code

  #2 opened 5 months ago by yuchen814
  3