jxbbb / TOD3Cap

[ECCV 2024] TOD3Cap: Towards 3D Dense Captioning in Outdoor Scenes
99 stars 5 forks source link

about training #5

Open hhhharold opened 3 months ago

hhhharold commented 3 months ago

After training on the complete nuscenes dataset, I cannot reproduce the performance of the paper. Is there any training tick?

SixCorePeach commented 1 week ago

these model just in release could be trained smoothly?