-
Notifications
You must be signed in to change notification settings - Fork 363
Added prompts for CrowS-Pairs-multilingual #748
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
Added prompts for CrowS-Pairs-multilingual #748
Conversation
I don't know that "Which sentence describes an unlikely event?" should necessarily include "event", so maybe in keeping with the prompt style of the rest of the set, something like "Which of the provided options is more realistic?" |
Furthermore, I'd refer to Albert's comment in this thread with respect to not having hard-coded responses. |
@thinkzink, thank you, I have integrated your suggestions! |
Now, both the French and English datasets have 7 prompts. |
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.
We may want to rename the dataset to remove username. Additionally we need the tasks to be considered original_tasks. @awebson may have an opinion?
promptsource/templates/oskarvanderwal/crows_pairs_multilingual/english/templates.yaml
Outdated
Show resolved
Hide resolved
promptsource/templates/oskarvanderwal/crows_pairs_multilingual/english/templates.yaml
Outdated
Show resolved
Hide resolved
* accelerate `get_infos` by caching the `DataseInfoDict`s * quality * consistency
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.
lgtm
We have added English and French prompts for the revised CrowS-pairs (multilingual) dataset: https://huggingface.co/datasets/oskarvanderwal/crows_pairs_multilingual. Some of the prompts are copied from this issue.
The major downside of the current prompts, is that the labels are hard-coded (always first or second choice).