Open utterances-bot opened 1 year ago
论文阅读： COLT5: Faster Long-Range Transformers with Conditional Computation 这篇文章是google research的，主要聚焦在这个方法怎么来让Transformer-based LLM能支持更长的context length，并做了一个64k context lenght的LLM，达到了GPT-4的两倍。但是，我个人认为，这
https://leoleoasd.me/2023/03/26/paper-reading-colt5-faster-long-range-transformers-with-conditional-computation/
test
leoleoasd
commented
1 year ago leoleoasd
commented
1 year ago
论文阅读： COLT5 + 运算优化真的只对长上下文有效吗？ | Leo
论文阅读： COLT5: Faster Long-Range Transformers with Conditional Computation 这篇文章是google research的，主要聚焦在这个方法怎么来让Transformer-based LLM能支持更长的context length，并做了一个64k context lenght的LLM，达到了GPT-4的两倍。但是，我个人认为，这
https://leoleoasd.me/2023/03/26/paper-reading-colt5-faster-long-range-transformers-with-conditional-computation/