-
Notifications
You must be signed in to change notification settings - Fork 1.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Restructuring training visualization and adding support for Weights & Biases Logger #664
base: develop
Are you sure you want to change the base?
Conversation
- VisualDL and W&B Logger inherit from a BaseLogger class - They have common functions `log_metrics`, `log_images` and more which internally call logger specific functions - Added functionality for users to use both VisualDL and W&B simultaneously - Added logging models to W&B as artifacts
…added documentation for WandbLogger
@LielinJiang @lijianshe02 I was wondering if it was possible to get some feedback |
@LielinJiang @LielinJiang I would love some feedback on this whenever you have the time |
Hi @LielinJiang @lijianshe02 It would be awesome if I could get this reviewed whenever possible |
Very glad to receive your pull request, we will review it as soon as possible |
Thx for your contribution. Could you add the function to PaddleMIX models https://github.com/PaddlePaddle/PaddleMIX/tree/develop |
|
This PR redesigns the logging system in GAN without changing any of the console logging. The proposed change is to have a BaseLogger class from which new loggers can inherit. This includes the VisualDL logger, the new Weights & Biases Logger and any new custom loggers that might be added later.
Using the W&B Logger would be really easy for users. The only requirement is to set
enable_wandb:True
in the config.yml file. To add any other wandb related arguments like project name, entity name and so on, the users can add the wandb key.For example:
This leads to all the training metrics being logged to W&B.
The dashboard:
The models at every model saving step and evaluation step are also logged to the W&B run with the appropriate tags and metadata.
The dashboard is available here