suzhenghang

About the focal loss layer

3

Hi @unsky , The performance in your experiment is amazing. By the way, did you only replace the SoftmaxWithLoss with the focal loss layer in RPN layer or in both...

Great work! Have there been any experimental results on integrating it into the Damo Text-to-Video system?

about deflicker

2

Do you have any good solutions for the flickering issue in generated videos?

About some warnings

1

I got the fllowing logs after running: python test_flops.py --config-file configs/RegNetX-4.0GF.ini，i wonder whether these warnings are normal? [INFO] Register count_convNd() for . [INFO] Register count_bn() for . [INFO] Register zero_ops()...

About VideoLDM

7

Do you have any knowledge of [VideoLDM](https://research.nvidia.com/labs/toronto-ai/VideoLDM/), and is it possible to integrate its algorithms to further enhance the capabilities of current models, such as generating longer videos?

[link](https://github.com/ExponentialML/Text-To-Video-Finetuning/blob/main/utils/dataset.py#L580)， device = torch.device("cuda" if torch.cuda.is_available() else "cpu") cached_latent = torch.load(self.cached_data_list[index], map_location=device) Otherwise, in multi-GPU distributed training, the first GPU may occupy excessive VRAM compared to the other GPUs.

如何清除历史上传的文档啊

2

如题

Do you have any good solutions for the flickering issue in generated videos?

多说话人效果

2

您有试过加了韵律特征后在多说话人上训练嘛？我这边多说话人训练效果没有单人训练的好，单人效果非常逼真