Integration of this repo with Weights and Biases(W&B). W&B allows to track machine learning experiments, visualize metrics, and share results. You can check it out here.
Motivations
This repository is really good for understanding the how various GAN flavors can be implemented. Thank you for this awesome resource.
To visualize the metrics one had to write his/her own code.
Not very convenient to do experiments involving hyperparameter tuning.
Not compatible with latest keras or tf.keras versions/ecosystem.
Contributions:
I tried instrumenting few scrips with W&B.
Make it work in tf.keras ecosystem. It can be changed back to keras ecosystem easily.
What this PR is about?
Integration of this repo with Weights and Biases(W&B). W&B allows to track machine learning experiments, visualize metrics, and share results. You can check it out here.
Motivations
kerasortf.kerasversions/ecosystem.Contributions:
tf.kerasecosystem. It can be changed back tokerasecosystem easily.Demonstration
Here is the notebook to demonstrate the use case: https://colab.research.google.com/drive/1u2gxV8tn42Vb_tmlGKa7Ww3xnEVcrxLB
(PS: You will need to signup with W&B. It's free and one can sign up using his/her GitHub account.)
Results
aae.pyhere.acgan.pyhere.bgan.pyhere.bigan.pyhere.ccgan.pyhere.cgan.pyhere.cogan.pyhere.dcgan.pyAdditional
Here is a report to support the brilliance of your work with W&B. The instrumentation that I did will allow the users of this repo to experiment with few configurations like different
latentdim. Link: https://app.wandb.ai/ayush-thakur/keras-gan/reports/Towards-Deep-Generative-Modeling-with-W%26B--Vmlldzo4MDI4Mw