from train_util import train_model
from dog_cat import load_datasets, prepare_datasets
from network import DistributionConvolutionModelKReluGradientMuRandom
from show_history import plot_history
if os.path.isdir("../model") == False:
os.mkdir("../model")
prepared = True
epochs = 300
device_name = "cuda"
learning_rate = 0.001
batch_size = 32
image_size = (100, 100)
model = DistributionConvolutionModelKReluGradientMuRandom(image_size=image_size)
device = torch.device(device_name)
model.to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=learning_rate)
if prepared == False:
prepare_datasets(image_size)
x_train, y_train, x_test, y_test = load_datasets()
print("Start Training")
model, history = train_model(model, criterion, optimizer, epochs, x_train, y_train, x_test, y_test, device, batch_size)
with open("../model/model_history.pkl", "wb") as f:
import torch
from distribution_3d import plot_distribution, plot_distribution_2d
from network import DistributionConvolutionModelKReluGradientMuRandom
import matplotlib.pyplot as plt
model = DistributionConvolutionModelKReluGradientMuRandom(image_size=(100,
100))
model.load_state_dict(
torch.load("../model/model_final.pth")["model_state_dict"])
layers = [[model.conv1_1, model.conv1_2], [model.conv2_1, model.conv2_2],
[model.conv3_1, model.conv3_2]]
shapes = [[4, 8], [8, 8], [8, 16]]
for i, layer in enumerate(layers):
for j, conv in enumerate(layer):
print("conv{}_{}.png".format(i + 1, j + 1))
distributions = conv.distribution.detach().numpy()
print(distributions.shape)
figure, graphs = plt.subplots(*shapes[i])
print(distributions.shape)
print(graphs.shape)
print(shapes[i][0] * shapes[i][1])
for graph, distribution in zip(
graphs.reshape(shapes[i][0] * shapes[i][1]), distributions):
graph.matshow(distribution)
graph.set_xticklabels([])
graph.set_yticklabels([])
graph.axis("off")
import torch
from distribution_3d import plot_distribution
from network import DistributionConvolutionModelKReluGradientMuRandom
import matplotlib.pyplot as plt
model = DistributionConvolutionModelKReluGradientMuRandom()
model.load_state_dict(torch.load("../model/model_final.pth"), strict=False)
distribution = model.conv3_1.distribution.detach().numpy()[2]
plot_distribution(distribution)