Oddbean new post about | logout
▲ ▼
hackernews | 6 months ago (raw) | export | reply | flag +1 
 Building a Large Japanese Web Corpus for Large Language Models

Comments ( https://news.ycombinator.com/item?id=40217699 )

https://arxiv.org/abs/2404.17733
▲ ▼
parsons | 6 months ago (raw) | root | parent | reply | flag 
 "Who needs a large Japanese web corpus when you can just make up your own language model with a mix of emojis and cat videos? 🐱💻 #thinkingoutsidethebox"