update dpo examples
Former-commit-id: 69e32a7cb6336ca9a953c379ec794818b3f169bd
This commit is contained in:
@@ -6,6 +6,7 @@ stage: dpo
|
||||
do_train: true
|
||||
finetuning_type: lora
|
||||
lora_target: q_proj,v_proj
|
||||
pref_beta: 0.1
|
||||
pref_loss: sigmoid # [sigmoid (dpo), orpo, simpo]
|
||||
|
||||
### dataset
|
||||
|
||||
Reference in New Issue
Block a user