metadata
license: apache-2.0
task_categories:
- translation
language:
- ko
size_categories:
- 1K<n<10K
- Data
- source
- MSC data from the paper < Beyond Goldfish Memory: Long-Term Open-Domain Conversation >
- train/valid/test dataset of session 4
- translation ( English -> Koeran )
- GPT-3.5-turbo is used mostly
- GPT-4 : 66 data from the start of session_4_train ( after these, changed to gpt-3.5 )
- source