Fix/hf datasets summarization default prompt#623
Conversation
| dataset_name = ( | ||
| self.dataset_name.split("-")[0].lower() if self.dataset_name else "sum_prompt" | ||
| ) | ||
|
|
There was a problem hiding this comment.
Does that refer to one of the key in default_user_prompt?? If yes, let's rename the key directly so that we don't have to handle edge cases
There was a problem hiding this comment.
@JulesBelveze
right know we were acessing the default_user_prompt based on dataset that we have in langtest
But, huggingface contain so many dataset so we will be using sum_prompt for all the huggingface dataset where task is summarization.
This is not the bug , this is is already in pr - #589
but due to some conflict I think the changes are not updated in the release/1.1.0 branch
There was a problem hiding this comment.
@Prikshit7766 Gotcha, could we rename it to something more self explanatory like "hf-dataset" or something? 😅
There was a problem hiding this comment.
@JulesBelveze
This prompt is like default summarization prompt not restricted to hf-dataset only.
if you have any suggestion we can do that
There was a problem hiding this comment.
Let's go for default-summarization-prompt
|
@Prikshit7766 In general it is good practice to either describe the bug you are trying to fix or link it to a GH issue. So that we know what you are trying to fix and can keep track of old bugs |
2 participants
Checklist:
pydanticfor typing when/where necessary.