search

TencentARC / SEED-Story

SEED-Story: Multimodal Long Story Generation with Large Language Model
https://arxiv.org/abs/2407.08683
Other
692 stars 54 forks source link

请问训练和测评是在三个数据集同时进行还是分开进行的呢 #21

Open shizhangyuya opened 3 weeks ago

shizhangyuya commented 3 weeks ago

数据集被分成了george，the land，rabbit三种，domain gap感觉还是比较大

AndysonYs commented 3 weeks ago

是分开进行的，但也可以放在一起训练，详见 #8