779ced0fabaaad57feedf544f452a69cf1c9baf6,examples/2d-distribution.py,,train,#Any#Any#,24

Before Change

                for k, metric in enumerate(gan.metrics.keys()):
                    if metric== "gradient_penalty":
                        print("--", gan.session.run(gan.metrics[metric]))
                        if math.isnan(gan.session.run(gan.metrics[metric])):
                            return None

        tf.reset_default_graph()
        gan.session.close()

    return sum_metrics

After Change

            if i % 300 == 0:
                for k, metric in enumerate(gan.metrics.keys()):
                    metric_value = gan.session.run(gan.metrics[metric])
                    print("--", metric,  metric_value)
                    if math.isnan(metric_value) or math.isinf(metric_value):
                        print("Breaking due to invalid metric")
                        return None

Instances