showlab / UniVTG

[ICCV2023] UniVTG: Towards Unified Video-Language Temporal Grounding

Home Page:https://arxiv.org/abs/2307.16715

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Loss parameters for downstream tasks

wenjiajia123 opened this issue Β· comments

Hi, I got bad results when training on moment retraival without loading pretrained model, I wonder that are the hyperparameters in 'qvhl_pretrain.sh' correct? πŸ™‚
b_loss_coef=10 g_loss_coef=0 eos_coef=0.1 f_loss_coef=10 s_loss_intra_coef=0 s_loss_inter_coef=0

Hi, this setting is not the optimal result. Please refer our paper or checkpoint config for parameters. :)

Thamk u! I have found that. 😊