torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
train_loader = DataLoader(get_dataset(dsize=args.size),
batch_size=args.batch_size,
shuffle=True)
test_loader = DataLoader(get_dataset(root="../data/test", dsize=args.size),
batch_size=args.batch_size,
shuffle=True)
model = VAE(args)
model.apply(weights_init())
if args.cuda:
model.cuda()
optimizer = optim.Adam(model.parameters(), lr=1e-4)
if args.rec_loss == "ssim":
rec_loss = SSIMLoss(method=args.ssim_method)
elif args.rec_loss == "bce":
rec_loss = nn.BCELoss(size_average=args.size_average)
elif args.rec_loss == "l1":
rec_loss = nn.L1Loss(size_average=args.size_average)
else:
rec_loss = None
kl_loss = KLLoss(size_average=args.size_average)
def loss_function(x, x_rec, mu, logvar):
def main():
CUDA = False
if torch.cuda.is_available():
CUDA = True
print('cuda available')
torch.backends.cudnn.benchmark = True
config = config_process(parser.parse_args())
print(config)
with open('pkl/task_1_train.pkl', 'rb') as f:
task_1_train = pkl.load(f)
# with open('pkl/task_1_test.pkl', 'rb') as g:
# task_1_test = pkl.load(g)
######################## task_1_testval.pkl
with open('pkl/task_1_test.pkl', 'rb') as g:
task_1_testval = pkl.load(g)
task_1_test = task_1_testval[:500]
task_1_val = task_1_testval[500:]
###################################### task 1 test + val
###### task 0:seen training data and unseen test data
examples, labels, class_map = image_load(config['class_file'],
config['image_label'])
###### task 0: seen test data
examples_0, labels_0, class_map_0 = image_load(config['class_file'],
config['test_seen_classes'])
datasets = split_byclass(config, examples, labels,
np.loadtxt(config['attributes_file']), class_map)
datasets_0 = split_byclass(config, examples_0, labels_0,
np.loadtxt(config['attributes_file']),
class_map)
print('load the task 0 train: {} the task 1 as test: {}'.format(
len(datasets[0][0]), len(datasets[0][1])))
print('load task 0 test data {}'.format(len(datasets_0[0][0])))
train_attr = F.normalize(datasets[0][3])
test_attr = F.normalize(datasets[0][4])
best_cfg = config
best_cfg['n_classes'] = datasets[0][3].size(0)
best_cfg['n_train_lbl'] = datasets[0][3].size(0)
best_cfg['n_test_lbl'] = datasets[0][4].size(0)
task_1_train_set = grab_data(best_cfg, task_1_train, datasets[0][2], True)
task_1_test_set = grab_data(best_cfg, task_1_test, datasets[0][2], False)
task_1_val_set = grab_data(best_cfg, task_1_val, datasets[0][2], False)
task_0_seen_test_set = grab_data(best_cfg, datasets_0[0][0],
datasets_0[0][2], False)
base_model = models.__dict__[config['arch']](pretrained=True)
if config['arch'].startswith('resnet'):
FE_model = nn.Sequential(*list(base_model.children())[:-1])
else:
print('untested')
raise NotImplementedError
print('load pretrained FE_model')
#######3 task id 'softmax'
FE_path = './ckpts/{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['softmax_method'], config['arch'],
config['task_id'], config['finetune'], 'checkpoint.pth')
FE_model.load_state_dict(torch.load(FE_path)['state_dict_FE'])
for name, para in FE_model.named_parameters():
para.requires_grad = False
vae = VAE(encoder_layer_sizes=config['encoder_layer_sizes'],
latent_size=config['latent_size'],
decoder_layer_sizes=config['decoder_layer_sizes'],
num_labels=config['num_labels'])
vae_path = './ckpts/{}_{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['method'], config['softmax_method'],
config['arch'], config['task_id'], config['finetune'], 'ckpt.pth')
vae.load_state_dict(torch.load(vae_path))
for name, para in vae.named_parameters():
para.requires_grad = False
FE_model.eval()
vae.eval()
# print(vae)
if CUDA:
FE_model = FE_model.cuda()
vae = vae.cuda()
#seen
task_1_real_train = get_prev_feat(FE_model, task_1_train_set, CUDA)
# task_0_real_val = get_prev_feat(FE_model, task_0_val_set, CUDA)
print('have got real trainval feats and labels')
print('...GENERATING fake features...')
task_0_fake = generate_syn_feature(150, vae, train_attr, config['syn_num'],
config)
train_X = torch.cat((task_0_fake[0].cuda(), task_1_real_train[0].cuda()))
train_Y = torch.cat(
(task_0_fake[1].cuda(), task_1_real_train[1].cuda() + 150))
# train_X = task_0_fake[0].cuda()
# train_Y = task_0_fake[1].cuda()
test_Dataset = PURE(train_X, train_Y)
test_dataloader = torch.utils.data.DataLoader(test_Dataset,
batch_size=256,
shuffle=True)
test_loss_net = nn.Linear(in_features=2048, out_features=200).cuda()
test_loss_net_optimizer = torch.optim.Adam(test_loss_net.parameters())
print('...TRAIN test set CLASSIFIER...')
# train_syn_val(test_loss_net,task_0_val_set, test_dataloader, test_loss_net_optimizer, 200)
best_loss_net = train_syn_val(config, FE_model, test_loss_net,
task_1_val_set, test_dataloader,
test_loss_net_optimizer, 200)
test_loss_net = copy.deepcopy(best_loss_net)
print('\n...TESTING... GZSL: 200 labels')
# test_loss_net = torch.load('hist_files/vae_time_2_doulbe_distill_loss_net.pth')
# torch.save(test_loss_net,'vae_time_2_loss_net.pth')
test_0_acc, test_0_top1, _, true_labels_0, pred_labels_0 = test(
config,
FE_model,
test_loss_net,
task_0_seen_test_set,
CUDA,
0,
eval_mode=0,
print_sign=1)
test_1_acc, test_1_top1, _, true_labels_1, pred_labels_1 = test(
config,
FE_model,
test_loss_net,
task_1_test_set,
CUDA,
1,
eval_mode=1,
print_sign=1)
H = 2 * test_0_acc * test_1_acc / (test_0_acc + test_1_acc)
print(H)
OM = (3 * test_0_acc + test_1_acc) / 4
print(OM)
if not os.path.exists('results'):
os.makedirs('results')
file_name = '{}_{}_{}_{}.txt'.format(config['dataset'], config['arch'],
config['method'], config['task_id'])
with open('results/' + file_name, 'a') as fp:
print(best_cfg, file=fp)
print(
'task B: {:.3f}, task A:{:.3f}, H= {:.3f}, OM= {:.3f} \n'.format(
test_0_acc, test_1_acc, H, OM),
file=fp)
def main():
file_name = 'data/CUB/attributes.txt'
f = open(file_name, 'r')
content = f.readlines()
atts = []
for item in content:
atts.append(item.strip().split(' ')[1])
CUDA = False
if torch.cuda.is_available():
CUDA = True
print('cuda available')
torch.backends.cudnn.benchmark = True
config = config_process(parser.parse_args())
print(config)
# pkl_name='./pkl/{}_{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
# config['dataset'], config['method'], config['softmax_method'],config['arch'],
# config['task_id'], config['finetune'], '.pkl')
#
# with open(pkl_name,'rb') as f:
# feat_dict=pkl.load(f)
with open('pkl/task_0_train.pkl', 'rb') as f:
task_0_train = pkl.load(f)
with open('pkl/task_1_train.pkl', 'rb') as f:
task_1_train = pkl.load(f)
###### task 0:seen training data and unseen test data
examples, labels, class_map = image_load(config['class_file'],
config['image_label'])
###### task 0: seen test data
examples_0, labels_0, class_map_0 = image_load(config['class_file'],
config['test_seen_classes'])
datasets = split_byclass(config, examples, labels,
np.loadtxt(config['attributes_file']), class_map)
datasets_0 = split_byclass(config, examples_0, labels_0,
np.loadtxt(config['attributes_file']),
class_map)
print('load the task 0 train: {} the task 1 as test: {}'.format(
len(datasets[0][0]), len(datasets[0][1])))
print('load task 0 test data {}'.format(len(datasets_0[0][0])))
classes_text_embedding = torch.eye(312, dtype=torch.float32)
test_attr = classes_text_embedding[:, :]
train_attr = F.normalize(datasets[0][3])
# test_attr=F.normalize(datasets[0][4])
best_cfg = config
best_cfg['n_classes'] = datasets[0][3].size(0)
best_cfg['n_train_lbl'] = datasets[0][3].size(0)
best_cfg['n_test_lbl'] = datasets[0][4].size(0)
task_0_train_set = grab_data(best_cfg, task_0_train, datasets[0][2], True)
task_1_train_set = grab_data(best_cfg, task_1_train, datasets[0][2], False)
base_model = models.__dict__[config['arch']](pretrained=True)
if config['arch'].startswith('resnet'):
FE_model = nn.Sequential(*list(base_model.children())[:-1])
else:
print('untested')
raise NotImplementedError
print('load pretrained FE_model')
FE_path = './ckpts/{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['softmax_method'], config['arch'],
config['task_id'], config['finetune'], 'checkpoint.pth')
FE_model.load_state_dict(torch.load(FE_path)['state_dict_FE'])
for name, para in FE_model.named_parameters():
para.requires_grad = False
vae = VAE(encoder_layer_sizes=config['encoder_layer_sizes'],
latent_size=config['latent_size'],
decoder_layer_sizes=config['decoder_layer_sizes'],
num_labels=config['num_labels'])
vae2 = VAE(encoder_layer_sizes=config['encoder_layer_sizes'],
latent_size=config['latent_size'],
decoder_layer_sizes=config['decoder_layer_sizes'],
num_labels=config['num_labels'])
vae_path = './ckpts/{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['method'], config['arch'], config['task_id'],
config['finetune'], 'ckpt.pth')
vae_path = './ckpts/{}_{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], 'vae', 'softmax_distill', config['arch'], 1,
config['finetune'], 'ckpt.pth')
vae2_path = './ckpts/{}_{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], 'vae_distill', 'softmax_distill', config['arch'], 1,
config['finetune'], 'ckpt.pth')
vae.load_state_dict(torch.load(vae_path))
vae2.load_state_dict(torch.load(vae2_path))
for name, para in vae.named_parameters():
para.requires_grad = False
for name, para in vae2.named_parameters():
para.requires_grad = False
if CUDA:
FE_model = FE_model.cuda()
vae = vae.cuda()
vae2 = vae2.cuda()
ATTR_NUM = 312
SYN_NUM = config['syn_num']
attr_feat, attr_lbl = generate_syn_feature(ATTR_NUM, vae, test_attr,
SYN_NUM, config)
attr_feat2, attr_lbl2 = generate_syn_feature(ATTR_NUM, vae2, test_attr,
SYN_NUM, config)
with open('attr_tsne_data/attr_vae_time_2_fe_distill.pkl', 'wb') as f:
pkl.dump(attr_feat, f)
with open('attr_tsne_data/attr_vae_time_2_double_distill.pkl', 'wb') as g:
pkl.dump(attr_feat2, g)
colors = cm.rainbow(np.linspace(0, 1, ATTR_NUM))
fig = plt.figure(figsize=(16, 9))
tsne = TSNE(n_components=2)
feat = torch.cat((attr_feat, attr_feat2))
tsne_results = tsne.fit_transform(feat)
color_ind = colors[attr_lbl]
ax = fig.add_subplot(1, 1, 1)
for i in range(ATTR_NUM):
ax.scatter(tsne_results[i * SYN_NUM:(i + 1) * SYN_NUM, 0],
tsne_results[i * SYN_NUM:(i + 1) * SYN_NUM, 1],
label=atts[i],
c=np.tile(colors[i].reshape(1, -1), (SYN_NUM, 1)),
s=20,
marker='X')
result = tsne_results[ATTR_NUM * SYN_NUM:, :]
for j in range(ATTR_NUM):
ax.scatter(result[j * SYN_NUM:(j + 1) * SYN_NUM, 0],
result[j * SYN_NUM:(j + 1) * SYN_NUM, 1],
label=atts[j],
c=np.tile(colors[j].reshape(1, -1), (SYN_NUM, 1)),
s=20,
marker='o')
# ax.scatter(tsne_results[:, 0], tsne_results[:, 1],c=color_ind, s=20, marker='X')
plt.legend()
plt.show()
def main():
CUDA = False
if torch.cuda.is_available():
CUDA = True
print('cuda available')
torch.backends.cudnn.benchmark = True
config = config_process(parser.parse_args())
print(config)
with open('pkl/task_1_train.pkl', 'rb') as f:
task_1_train = pkl.load(f)
with open('pkl/task_1_test.pkl', 'rb') as g:
task_1_test = pkl.load(g)
###### task 0:seen training data and unseen test data
examples, labels, class_map = image_load(config['class_file'],
config['image_label'])
###### task 0: seen test data
examples_0, labels_0, class_map_0 = image_load(config['class_file'],
config['test_seen_classes'])
datasets = split_byclass(config, examples, labels,
np.loadtxt(config['attributes_file']), class_map)
datasets_0 = split_byclass(config, examples_0, labels_0,
np.loadtxt(config['attributes_file']),
class_map)
print('load the task 0 train: {} the task 1 as test: {}'.format(
len(datasets[0][0]), len(datasets[0][1])))
print('load task 0 test data {}'.format(len(datasets_0[0][0])))
test_attr = F.normalize(datasets[0][4])
best_cfg = config
best_cfg['n_classes'] = datasets[0][3].size(0)
best_cfg['n_train_lbl'] = datasets[0][3].size(0)
best_cfg['n_test_lbl'] = datasets[0][4].size(0)
task_1_train_set = grab_data(best_cfg, task_1_train, datasets[0][2], True)
# task_1_test_set = grab_data(best_cfg, task_1_test, datasets[0][2], False)
base_model = models.__dict__[config['arch']](pretrained=False)
if config['arch'].startswith('resnet'):
FE_model = nn.Sequential(*list(base_model.children())[:-1])
else:
print('untested')
raise NotImplementedError
###### if finetune == False,
print('load pretrained FE_model')
FE_path = './ckpts/{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['softmax_method'], config['arch'],
config['task_id'], config['finetune'], 'checkpoint.pth')
FE_model.load_state_dict(torch.load(FE_path)['state_dict_FE'])
for name, para in FE_model.named_parameters():
para.requires_grad = False
vae = VAE(encoder_layer_sizes=config['encoder_layer_sizes'],
latent_size=config['latent_size'],
decoder_layer_sizes=config['decoder_layer_sizes'],
num_labels=config['num_labels'])
vae_path = './ckpts/{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], 'vae', config['arch'], 0, config['finetune'],
'ckpt.pth')
vae.load_state_dict(torch.load(vae_path))
print(vae)
if CUDA:
FE_model = FE_model.cuda()
vae = vae.cuda()
FE_model.eval()
optimizer = torch.optim.Adam(vae.parameters(), lr=config['lr'])
scheduler = torch.optim.lr_scheduler.StepLR(optimizer,
config['step'],
gamma=0.1,
last_epoch=-1)
criterion = loss_fn
print('have got real trainval feats and labels')
for epoch in range(config['epoch']):
print('\n epoch: %d' % epoch)
print('...TRAIN...')
print_learning_rate(optimizer)
### train_attr--->test_attr, task_0_train_set---> task_1_train_set
train(epoch, FE_model, vae, task_1_train_set, optimizer, criterion,
test_attr, CUDA)
scheduler.step()
vae_ckpt_name = './ckpts/{}_{}_{}_{}_task_id_{}_finetune_{}_{}'.format(
config['dataset'], config['method'], config['softmax_method'],
config['arch'], config['task_id'], config['finetune'], 'ckpt.pth')
torch.save(vae.state_dict(), vae_ckpt_name)
default=False,
metavar='UT',
help='load pretrained model (default: False)')
args = parser.parse_args()
batch_loader = BatchLoader()
parameters = Parameters(batch_loader.vocab_size)
vae = VAE(parameters.vocab_size, parameters.embed_size,
parameters.latent_size, parameters.decoder_rnn_size,
parameters.decoder_rnn_num_layers)
if args.use_trained:
vae.load_state_dict(t.load('trained_VAE'))
if args.use_cuda:
vae = vae.cuda()
optimizer = Adam(vae.parameters(), args.learning_rate)
for iteration in range(args.num_iterations):
'''Train step'''
input, decoder_input, target = batch_loader.next_batch(
args.batch_size, 'train', args.use_cuda)
target = target.view(-1)
logits, aux_logits, kld = vae(args.dropout, input, decoder_input)
logits = logits.view(-1, batch_loader.vocab_size)
cross_entropy = F.cross_entropy(logits, target, size_average=False)
aux_logits = aux_logits.view(-1, batch_loader.vocab_size)
