out_dir = './out/gan_{}'.format(datetime.now()) out_dir = out_dir.replace(" ", "_") print(out_dir) if not os.path.exists(out_dir): os.makedirs(out_dir) shutil.copyfile(sys.argv[0], out_dir + '/training_script.py') sys.stdout = mutil.Logger(out_dir) gpu = 0 torch.cuda.set_device(gpu) mb_size = 600 # mini-batch_size mode_num = 2 distance = 10 data = data_prepare.Data_2D_Circle(mb_size, mode_num, distance) Z_dim = 4 X_dim = 2 h_dim = 128 c_dim = mode_num * mode_num cnt = 0 num = '0' # else: # print("you have already creat one.") # exit(1) G = model.G_Net(Z_dim + c_dim, X_dim, h_dim).cuda() D = model.D_Net_w(X_dim, 1, h_dim).cuda()
def add_mode(mode_num ,distance): # mode_num = mn+1 # distance = distance * 1.5 return data_prepare.Data_2D_Circle(mb_size,mode_num,distance, noise_variance=0.5)
torch.cuda.set_device(gpu) mb_size = 96 # mini-batch_size # mode_num = 2 sample_point = 10000 # distance = 10 # start_points = np.array([[0,0],[0,1],[0,2]]) # end_points = np.array([[1,0],[1,1],[1,2]]) start_points = np.array([[0, 0]]) end_points = np.array([[1, 0]]) Z_dim = 2 X_dim = 2 h_dim = 16 # data = data_prepare.Straight_Line(90, start_points, end_points, type=1) data = data_prepare.Data_2D_Circle(mb_size, R=2) data_draw_m = data_prepare.Data_2D_Circle(8, R=2) data_draw = data_draw_m.batch_next() z_draw = Variable(torch.randn(sample_point, Z_dim)).cuda() # c_dim = mode_num * mode_num cnt = 0 num = '0' # else: # print("you have already creat one.") # exit(1) grid_num = 100 top_line = 3
out_dir = './out/gan_add_mode_{}'.format(datetime.now()) out_dir = out_dir.replace(" ", "_") print(out_dir) if not os.path.exists(out_dir): os.makedirs(out_dir) shutil.copyfile(sys.argv[0], out_dir + '/training_script.py') sys.stdout = mutil.Logger(out_dir) gpu = 6 torch.cuda.set_device(gpu) mb_size = 600 # mini-batch_size mode_num = 2 distance = 5 data = data_prepare.Data_2D_Circle(mb_size, mode_num, distance, noise_variance=0.5) Z_dim = 2 X_dim = 2 h_dim = 128 c_dim = mode_num * mode_num cnt = 0 num = '0' # else: # print("you have already creat one.") # exit(1) G = model.G_Net(Z_dim , X_dim, h_dim).cuda() D = model.D_Net(X_dim , 1, h_dim).cuda()