Latest version RNN-T operation issues

[llleohk]

When I downloaded the latest wenet code and ran it, I reported an error：

Traceback (most recent call last): File "wenet/bin/train.py", line 176, in main() File "/data/.conda/envs/wenet/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 345, in wrapper return f(*args, kwargs) File "wenet/bin/train.py", line 142, in main executor.train(model, optimizer, scheduler, train_data_loader, File "/data/wenet-moe/wenet-3.0.0/wenet/utils/executor.py", line 74, in train info_dict = batch_forward(model, batch_dict, scaler, File "/data/wenet-moe/wenet-3.0.0/wenet/utils/train_utils.py", line 471, in batch_forward loss_dict = model(batch, device) File "/data/.conda/envs/wenet/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, *kwargs) File "/data/.conda/envs/wenet/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 1008, in forward output = self._run_ddp_forward(inputs, kwargs) File "/data/.conda/envs/wenet/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 971, in _run_ddp_forward return module_to_run(*inputs, *kwargs) File "/data/.conda/envs/wenet/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(input, *kwargs) File "/data/wenet-moe/wenet-3.0.0/wenet/transducer/transducer.py", line 139, in forward loss = loss + self.ctc_weight loss_ctc.sum() AttributeError: 'tuple' object has no attribute 'sum'

How should I handle it？

[kobenaxie]

CTC model returns a tuple now, can be fixed by https://github.com/wenet-e2e/wenet/pull/2327

[llleohk]

CTC model returns a tuple now, can be fixed by #2327

thank you very much