zyzisyz / mfa_conformer Goto Github PK

Hi,

Thank you for publishing your code. However, I am unable to reproduce your result (My EER: 1.24%). I believe using different torch versions lead to the inconsistent results. Therefore, may I ask which torch version you are using in this model.

Once again, thank you for sharing your work!

@zyzisyz any plan to test this dataset?

Traceback (most recent call last):
File "main.py", line 256, in
cli_main()
File "main.py", line 247, in cli_main
log_every_n_steps=25,
File "/mnt/workspace/workgroup/yfchen/anaconda3/envs/py37/lib/python3.7/site-packages/pytorch_lightning/utilities/argparse.py", line 345, in insert_env_defaults
return fn(self, **kwargs)
File "/mnt/workspace/workgroup/yfchen/anaconda3/envs/py37/lib/python3.7/site-packages/pytorch_lightning/trainer/trainer.py", line 450, in init
plugins=plugins,
File "/mnt/workspace/workgroup/yfchen/anaconda3/envs/py37/lib/python3.7/site-packages/pytorch_lightning/trainer/connectors/accelerator_connector.py", line 200, in init
sync_batchnorm=sync_batchnorm,
File "/mnt/workspace/workgroup/yfchen/anaconda3/envs/py37/lib/python3.7/site-packages/pytorch_lightning/trainer/connectors/accelerator_connector.py", line 328, in _check_config_and_set_final_flags
f"Found invalid type for plugin {plugin}. Expected one of: PrecisionPlugin, "
pytorch_lightning.utilities.exceptions.MisconfigurationException: Found invalid type for plugin <pytorch_lightning.plugins.training_type.ddp.DDPPlugin object at 0x7f9dc58201d0>. Expected one of: PrecisionPlugin, CheckpointIO, ClusterEnviroment, or LayerSync.

Can you share the pre-trained model checkpoint for conformer model?

Sorry, Can you provide the SITW.Dev and SITW.Eval trials used in the paper?

I found this kind of error when running the code, has anyone encountered it and how to solve it?

Hi, thank you to share your official code of mfa_conformer.

I also implement your score successfully, but I have one of question.

When I run your code, the total number of data loader is 10,912.

Actually, the total number of Vox1&2's utterances is 1,240,651, and batch_size is 200. Then, the estimated iteration numbers should be 6203.

What is 10,912 - 6,203 = 4709 iterations ?

Thanks

Hi, thanks for your excellent work. May I ask how you calculate the real-time factor? Do you have the implementation in your code? thank you very much.

Regarding multi-gpu training, is your code trained on a single card or multiple cards?

Thanks for your great work! I kewn model code based on wenet, could you tell what are used pipline ?