wespeaker: Get stuck in load a batch

Hi, dear.

I am following the settings of the original script, using the shards data type, I want to know, is your experiment using SSD? Is it because of the fuck mechanical hard disk that it has been stuck in load a batch? It is not very slow, but it is always stuck there forever.

I have a record about it

`

    print("before get a batch")
    for i, batch in enumerate(dataloader):
        print("get a batch")
        utts = batch['key']
        targets = batch['label']
        features = batch['feat']

`

after, … of, there is no after.