Skip to content

Align KTO with DPO: Remove enforcement of causal language models #5701

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
albertvillanova merged 1 commit into from
May 5, 2026
+0 −7
Merged

Align KTO with DPO: Remove enforcement of causal language models #5701

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 0 additions & 7 deletions trl/experimental/kto/kto_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -324,13 +324,6 @@ def __init__(
if param.requires_grad:
param.data = param.data.to(torch.bfloat16)

# KTO only supports causal language models, not encoder-decoder models
if model is not None and hasattr(model.config, "is_encoder_decoder") and model.config.is_encoder_decoder:
raise ValueError(
"KTO only supports causal language models. Encoder-decoder models are not supported. "
"Please use a causal LM (e.g., GPT, Llama, Mistral) instead of an encoder-decoder model (e.g., T5, BART)."
)

if args.max_length is None:
logger.warning(
"When using DataCollatorForUnpairedPreference, you should set `max_length` in the KTOTrainer's init"
Expand Down
Loading