Validation now included during training

deepmagd / shenaniGAN

Work in progress

MIT License

1 stars 0 forks source link

Validation now included during training #31

Closed alecokas closed 4 years ago

alecokas commented 4 years ago

We now pass the validation loader in with the train loader so that after each epoch we can run validation. We log the log the validation losses.
I also added the text embedding as a parameter in the dataset object

Review this after PR #28 as this branch splits off from there. Addresses issue #29

alecokas commented 4 years ago

I have now also added parameters which indicate how often the model should be saved during training and after how many epochs we should start tracking the best model. This PR also deals with #30 now.