-
Notifications
You must be signed in to change notification settings - Fork 14
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.
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
Penalty rewards #52
Penalty rewards #52
Conversation
With the new prompting style you receive much worse ratings from the RLHF model. I whould think the RLHF model doesn't do well, when the question is at the start of the prompt |
…sor/text-prompting into features/penalty_rewards
…sor/text-prompting into features/penalty_rewards
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please make penalty_scale_factor
in prompting/validators/criteria.py
a gaussian as i showed in the plot.
This PR seeks to propose a penalty mechanism with:
- Penalty rewards: A new family of function models that will act as penalty functions accordingly to the established definition of the function. This PR also introduces a task-criteria schema that enable criteria assignment to a given task through prompt definition and code validation.
Tasks done so far:
SummaryTask
,QuestionGenerationTask
andQuestionAnswerTask
MatchLengthCriteria
with support to characters, words, sentences and paragraphsKeywordMatch
penalty functionSentenceMatch
penalty functionTODO (Once current design is approved):
Wandb run with penalty functions implemented: