README.md

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 )