Please upload saved model about keras-gan HOT 5 CLOSED

I actually have applied this saved models function to most of your GANs implementation. My request was to upload saved hdf5 file of your implementation as training models are computationally expensive and people who are interested in results can use those models. But anyways I have implemented these models and can upload my hdf5 files if you would allow me to.
P.S. your implementation helped me a lot in understanding GANs.

from keras-gan.

@tejasvx You can train the models in less then a couple of hours tops, create a Google Cloud Account which gives you free credits to work with.

from keras-gan.

GCP free credits does not contain GPU instances but I trained the model on my laptop and it worked great.
Feel like I should close this issue anyways.

from keras-gan.

Some GAN implementations have a method save_model. This is from acgan/acgan.py:

    def save_model(self):

        def save(model, model_name):
            model_path = "acgan/saved_model/%s.json" % model_name
            weights_path = "acgan/saved_model/%s_weights.hdf5" % model_name
            options = {"file_arch": model_path, 
                        "file_weight": weights_path}
            json_string = model.to_json()
            open(options['file_arch'], 'w').write(json_string)
            model.save_weights(options['file_weight'])

        save(self.generator, "mnist_acgan_generator")
        save(self.discriminator, "mnist_acgan_discriminator")
        save(self.combined, "mnist_acgan_adversarial")

I have not implemented this method for all GAN varieties, but slight modifications to this method should help you save the model you want to train.

from keras-gan.

#70

from keras-gan.