Got crashed after the first epoch training on windows 11. about adaseq HOT 4 CLOSED

Please set workers_per_gpu: 1 to 0 in examples/bert_crf/configs/resume.yaml

from adaseq.

From modelscope 1.0.0 to 1.1.0, all crashed with the same messages.
That's a tricky one. 😿

I'm afraid we can't use it on windows, until it gets fixed.

from adaseq.

Please set workers_per_gpu: 1 to 0 in examples/bert_crf/configs/resume.yaml

Thanks, we also located the problem.
By disabling the multi-process loading of dataloader, i.e., set workers_per_gpu: 0, we are fine on windows.

We will change the default configurations.

from adaseq.

Fixed

from adaseq.