How to continue the training after an interruption?

[luoshuiyue]

Could you please provide the code to continue the training after the training is interrupted?

[theAdityaRamesh]

• python train.py --cfg configs/human_nerf/zju_mocap/{your_subj}/{your_yaml_file_name.yaml} resume True in the cmdline
• Training Resumes from previous checkpoint

[luoshuiyue]

Thanks for your help. But I don't know what to write in yaml file. If the data set is provided by the author, then the yaml file should be the same, so how exactly to train resuming from previous checkpoint? Whether to add a command argument or write another pyhton file, or change a python file?

• python train.py --cfg configs/human_nerf/zju_mocap/{your_subj}/{your_yaml_file_name.yaml} resume True in the cmdline
• Training Resumes from previous checkpoint

[theAdityaRamesh]

1. You don't have to write anything in the yaml file.
2. By your yaml file i meant whatever yaml file you are using to train eg : adventure.yaml or single_gpu.yaml
3. You have to go to the folder where train.py is and type the above in the command line.
4. The training resumes from the last check point.

[luoshuiyue]

I tried just like the author provided the command and you provided above, but it will remind me that i have to remove the existing log file and reproduce the training. If I enter 'no' it will exit. Are you saying that even though the training is interrupted, running the same training command again will automatically resume the training from the break point?

1. You don't have to write anything in the yaml file.
2. By your yaml file i meant whatever yaml file you are using to train eg : adventure.yaml or single_gpu.yaml
3. You have to go to the folder where train.py is and type the above in the command line.
4. The training resumes from the last check point.

[theAdityaRamesh]

Refer issue #20

[luoshuiyue]

Refer issue #20

Thank you very much!