static void Main(string[] args)
{
KernelManager.Initialize();
KernelManager.GPUMode = true;
//var superResolution = new GAN();//new ReversibleAutoencoder(); //new ConvSuperResolution();
new ConvSuperResolution().Train();
//new GradientChecking().Check();
}
static void Main(string[] args)
{
KernelManager.Initialize();
var tags = new string[]
{
//"emilia_(re:zero)",
//"katou_megumi",
//"tokisaki_kurumi",
//"sawamura_spencer_eriri",
//"kasumigaoka_utaha",
//"zero_two_(darling_in_the_franxx)",
//"nishikino_maki",
//"souryuu_asuka_langley",
//"shiba_miyuki",
//"akemi_homura",
//"kizuna_ai",
//"euryale",
"osakabe-hime_(fate/grand_order)",
"momo_velia_deviluke",
"semiramis_(fate)",
"altera_(fate)",
"takao_(aoki_hagane_no_arpeggio)",
"medb_(fate)_(all)",
"meltlilith",
"yuzuriha_inori",
//"redjuice"
};
var globalTags = new string[]
{
"1girl",
"solo",
"-*boy*",
"-large_breasts",
"-video",
};
BooruDatasetBuilder datasetBuilder = new BooruDatasetBuilder();
for (int i = 0; i < globalTags.Length; i++)
{
datasetBuilder.AddGlobalTag(globalTags[i]);
}
for (int i = 0; i < tags.Length; i++)
{
datasetBuilder.AddLocalTag(tags[i]);
}
datasetBuilder.Download(500, @"I:\Datasets\Gelbooru");
//var inputDataset = datasetBuilder.GetDataset(@"I:\Datasets\Gelbooru", @"I:\Datasets\Gelbooru_SMALL", Side, 250);
/*
* var classifier = new NeuralNetworkBuilder(Side * Side * 3)
* .WeightInitializer(new UniformWeightInitializer(0, 0))
* .LossFunction<Quadratic>()
* .AddConv(3, 3, 1, 0, Side, 3)
* .AddActivation<ReLU>()
* .AddPooling(2, 2, 3)
* .AddConv(3, 10, 1, 0, Side / 2, 3)
* .AddActivation<ReLU>()
* .AddPooling(2, 2, 10)
* .AddFC(4096)
* .AddActivation<LeakyReLU>()
* .AddFC(1024)
* .AddActivation<ReLU>()
* .AddFC(512)
* .AddActivation<ReLU>()
* .AddFC(256)
* .AddActivation<ReLU>()
* .AddFC(64)
* .AddActivation<ReLU>()
* .AddFC(16)
* .AddActivation<ReLU>()
* .AddFC(8)
* .AddActivation<ReLU>()
* .AddFC(8)
* .AddActivation<ReLU>()
* .AddFC(8)
* .AddActivation<ReLU>()
* .AddFC(tags.Length)
* .AddActivation<Sigmoid>()
* .Build();
*
* var trainer = new ClassifierTrainer("Anime Classifier", tags, classifier);
* trainer.SetDataset(inputDataset);
*
* LearningManager.Show(trainer);*/
}