search

MCG-NKU / E2FGVI

Official code for "Towards An End-to-End Framework for Flow-Guided Video Inpainting" (CVPR2022)
Other
1.04k stars 100 forks source link

about the loss #38

Open Davidcoach opened 2 years ago

Davidcoach commented 2 years ago

Hello, when I try to train your e2fgvi_hq model on the youtube-vos datasets, after 400k iterations the model will always collapse, and the flow loss will became nan. Have you ever face this problem? Looking forward to your answer!

Paper99 commented 2 years ago

Hello, you could refer to this issue #7.

MasterHow commented 2 years ago

same question，the training is not stable.