[Help]: NaturalSpeech2 training and data preprocess issue

[CreepJoye]

Hello,I would like to run sh egs/tts/NaturalSpeech2/run_train.sh,and it reports :
usage: accelerate <command> [<args>]
Accelerate CLI tool: error: argument {config,estimate-memory,env,launch,tpu-config,test}: invalid choice: /home/lqb/project/Amphion/bins/tts/train.py' (choose from 'config', 'estimate-memory', 'env', 'launch', 'tpu-config', 'test')
How should I handle this issue?
thank you for your patient explanation!

[CreepJoye]

1. When I follow other Amphion TTS recipes for the data processing,it seems works but not totally.I get these files in egs/tts/NaturalSpeech2/data_processed/libritts,but no content features.
   
   
2. How should I set the path of the "metadata_dir" in NaturalSpeech2/exp_config.json/"preprocess"?I don't know what is "metadata" and it seems like I get no "metadata" since I have created a 'metadata' folder and set the path of 'metadata' to this folder, but it is empty.
3. I complete preprocess.py, and also obtain train.json and valid.json.Then I start to run train.py,and it reports
   

Your responses are greatly appreciated！！！