language: | |
- ko | |
tags: | |
- text generation | |
- pytorch | |
- causal-lm | |
license: apache-2.0 | |
datasets: | |
- lcw99/wikipedia-korean-20221001 | |
- heegyu/namuwiki-extracted | |
- cc100 | |
- oscar | |
# gpt-neo-1.3B Korean version | |
PPL on Oscar Korean text dataset = 46.0 |