def __init__(self, db_path, db_type, kmeans_file): # initialize the renderedImages dataset first super().__init__(db_path, db_type, 'quaternion') # add the kmeans part self.kmeans = pickle.load(open(kmeans_file, 'rb')) self.num_clusters = self.kmeans.n_clusters self.kmeans.cluster_centers_ = convert_dictionary(self.kmeans.cluster_centers_)
def __init__(self, alpha, kmeans_file, my_loss):
super().__init__()
self.alpha = alpha
kmeans = pickle.load(open(kmeans_file, 'rb'))
self.cluster_centers = Variable(
torch.from_numpy(convert_dictionary(
kmeans.cluster_centers_)).float()).cuda()
self.n_clusters = kmeans.n_clusters
self.my_loss = my_loss
self.kl = nn.KLDivLoss().cuda()
def __init__(self, alpha, kmeans_file, my_loss=None):
super().__init__()
self.alpha = alpha
kmeans = pickle.load(open(kmeans_file, 'rb'))
self.cluster_centers_ = Variable(
torch.from_numpy(convert_dictionary(
kmeans.cluster_centers_)).float()).cuda()
if my_loss is None:
self.mse = nn.MSELoss().cuda()
else:
self.mse = my_loss
self.ce = nn.CrossEntropyLoss().cuda()
args = parser.parse_args()
print(args)
# assign GPU
os.environ['CUDA_VISIBLE_DEVICES'] = args.gpu_id
# save stuff here
results_file = os.path.join('results', args.save_str)
model_file = os.path.join('models', args.save_str + '.tar')
plots_file = os.path.join('plots', args.save_str)
log_dir = os.path.join('logs', args.save_str)
# kmeans data
kmeans_file = 'data/kmeans_dictionary_axis_angle_' + str(
args.dict_size) + '.pkl'
kmeans = pickle.load(open(kmeans_file, 'rb'))
kmeans_dict = convert_dictionary(kmeans.cluster_centers_)
num_clusters = kmeans.n_clusters
# relevant variables
ndim = 4
num_classes = len(classes)
if not args.multires:
criterion = RelaXedProbabilisticLossQ(1.0, kmeans_file,
geodesic_loss(reduce=False).cuda())
else:
criterion = RelaXedProbabilisticMultiresLossQ(
1.0, kmeans_file,
geodesic_loss(reduce=False).cuda())
# DATA