How to train

[see2run]

Hello, can anyone help me?

I'm processing a dataset using preprocess.py and listing its subfolders into a txt file.

Then, the output that appears is only:

use_cuda: True total trainable params 65054464 Training From Scratch !!! Starting Epoch: 0 True

for the True output there I'm trying to make sure of the dataloader with if train_data_loader: print(True) else: print(False)

[see2run]

Because after I checked, this loop process isn't running but also not showing any errors:

for step, (x, mel, y) in enumerate(train_data_loader): st = time() print("running") model.train() optimizer.zero_grad()

where the print("running") is not displayed

[LinYi-JCY]

Have you solved the problem?

[see2run]

Have you solved the problem?

yeah

[LinYi-JCY]

Can you provide some solutions? I also encountered this problem, thank you very much!🙏

[see2run]

Can you provide some solutions? I also encountered this problem, thank you very much!🙏

What I did was just fix the train syncnet and wav2lip scripts in the Dataset class. Pay attention to whether the path for obtaining the list file is correct and find out if there is any variable with a None value in that class. Then, at the very least, use more than 1000 training data and more than 100 test data (if you want to test whether the script works or not). If there is anything else you want to ask, feel free to ask

[LinYi-JCY]

What I did was just fix the train syncnet and wav2lip scripts in the Dataset class. Pay attention to whether the path for obtaining the list file is correct and find out if there is any variable with a None value in that class. Then, at the very least, use more than 1000 training data and more than 100 test data (if you want to test whether the script works or not). If there is anything else you want to ask, feel free to ask

Ok, I will try this solution. Thank you again!