Closed alvarobartt closed 5 months ago
This PR fixes the test_orpo_trainer_demo.py as it was overriding the value of sample["chosen"] leading to a bad formatting within the prompt.
test_orpo_trainer_demo.py
sample["chosen"]
prompt
Additionally, this PR also applies the following minor changes related to formatting and readability:
if __name__ == "__main__"
os
transformers
Thank you for the fix and clarification @alvarobartt😃
Description
This PR fixes the
test_orpo_trainer_demo.py
as it was overriding the value ofsample["chosen"]
leading to a bad formatting within theprompt
.Additionally, this PR also applies the following minor changes related to formatting and readability:
if __name__ == "__main__"
os
import and duplicatedtransformers
imports