Does Rave Training stop by itself or should I do it by myself #130
-
|
Dear friends, I am training a dataset made of percussions. I just reached 2.6 million points in 3 days of continuos processing. I do not remember if Rave stops by itself at a certain point, or if I should put an end to the training and to export the model. Any suggestion would be of great help. Thanks to everyone. Federico Placidi |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 5 replies
-
|
If your size is set to default, train_rave.py will stop by itself at 3 Million steps. |
Beta Was this translation helpful? Give feedback.
-
|
What are "steps" vs. Epoch? Where are the steps shown in the collab? Thanks for help. |
Beta Was this translation helpful? Give feedback.
-
|
See #90 |
Beta Was this translation helpful? Give feedback.
-
|
regarding colab, you can monitor everything using tensorboard. A guide on how to interpret tensorboard metrics for a rave training is available here https://github.com/acids-ircam/RAVE/blob/master/docs/tensorboard_guide.md |
Beta Was this translation helpful? Give feedback.
-
Is it possible to safely stop and export the training before that 3 million steps? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, you can just kill the process when you are satisfied with the output quality (you can listen to it over tensorboard. |
Beta Was this translation helpful? Give feedback.
If your size is set to default, train_rave.py will stop by itself at 3 Million steps.