run train.py meets ValueError: `Dataloader` returned 0 length.

Hello!

I am studying the recipe of /asteroid-master/egs/dns_challenge/ and trying to run it for practice. But I met error and I failed to train the model. I run it in my laptop in Linux, I installed pytorch_lightning version 0.9.0 from PyPI. But when I run /asteroid-master/egs/dns_challenge/baseline/run.sh at Stage 3 : Train, it will throw error at pytorch_lightning. The error is as shown below. Train_error_log.txt For more detail please find the attached Train_error_log.txt.

---

Traceback (most recent call last): File “train.py”, line 121, in <module> main(arg_dic) File “train.py”, line 94, in main trainer.fit(system) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/states.py”, line 48, in wrapped_fn result = fn(self, *args, **kwargs) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/trainer.py”, line 1084, in fit results = self.accelerator_backend.train(model) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/accelerators/cpu_backend.py”, line 39, in train results = self.trainer.run_pretrain_routine(model) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/trainer.py”, line 1224, in run_pretrain_routine self._run_sanity_check(ref_model, model) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/trainer.py”, line 1249, in _run_sanity_check self.reset_val_dataloader(ref_model) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/data_loading.py”, line 337, in reset_val_dataloader self.num_val_batches, self.val_dataloaders = self._reset_eval_dataloader(model, ‘val’) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/data_loading.py”, line 299, in _reset_eval_dataloader num_batches = len(dataloader) if _has_len(dataloader) else float(‘inf’) File “/home/elen/anaconda3/lib/python3.8/site-packages/pytorch_lightning-0.9.1rc4-py3.8.egg/pytorch_lightning/trainer/data_loading.py”, line 70, in _has_len raise ValueError(‘Dataloader returned 0 length.’ ValueError: Dataloader returned 0 length. Please make sure that your Dataloader at least returns 1 batch

---

I have successfully run it for Stage 1 and 2, and the file_infos.json already created at location ./data. I don’t know what cause this Dataloader to return 0 length. Can you help me to point out what maybe the cause? How can I solve this problem? Thanks in advance!

Best Regards, Elen

[elenzheng]

I downloaded the default master Branch code just last week. From the Changelog I found its version is [0.3.3] - 2020-08-25. Now after removing the argument ‘train’ from common_step() in the code of /asteroid/engine/system.py, I can successfully run the Stage 3 to train the model.

[mpariente]

• Because you don’t have many examples.
• What is your asteroid version?

Please, try debugging before posting the issue, if you solve your problems, you’ll be better after it.