Closed MuroriM closed 3 years ago
I've fixed how the dialogue instances are stored. Each source instance is now a list of strings in the form of '["speaker1 : utter..", "speaker2 : utter..."]
LGTM. @MuroriM Can you confirm that the tests are passing?
It's passing the tests. I also modified to check for list[str] instances on source if it's a dialogue dataset
Awesome, merging this now
Add SAMsum dataset:
Does minor edits to fix datasets naming convention on Scisummnet