-
Notifications
You must be signed in to change notification settings - Fork 418
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
Add new setting: prompt_column_separator #724
Conversation
Docs LGTM, thank you! |
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.
THX - overall looks good!
A little feedback:
- Shall we add a different default separator, like \n###\n or similar?
- I guess it is not possible to add a nesting that only triggers if multiple values are selected? Possible probably, but some work, not sure if worth.
- Otherwise I would suggest to add it to the tokenizer section though. It is a bit dominant on top.
@@ -40,6 +40,7 @@ class ConfigNLPCausalLMDataset(DefaultConfig): | |||
|
|||
system_column: str = "system" | |||
prompt_column: Tuple[str, ...] = ("instruction", "input") | |||
prompt_column_separator: str = "" |
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.
Should this be an empty string by default?
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.
It was double space before, which is hard to see in the textbox and also a bit unusual. Decided to change it to double newlines now. Will push the changes in a bit. Wdyt?
After some thought, I decided to use
I also thought about it, but couldn't find an easy way. It would also force us to use a trigger for the columns dropdown.
From a logical standpoint, I believe it should be a dataset setting. |
closes #483