3 ms·
Nothing new here, just yet another wrapper around the current language models, and weak ones at that.
by behnamoh 3y ago
Nothing new here, just yet another wrapper around the current language models, and weak ones at that.
- rolisz 3y agoFlan-T5 models perform really well, considering their size. In my experiments and in some recent papers, they are very close to 13b Llama for example: https://twitter.com/YiTayML/status/1668302949276356609 https://twitter.com/YiTayML/status/1668302949276356609
- yieldcrv 3y agoits like the bar keeps getting lower and lower