def main():
import argparse
parser = argparse.ArgumentParser(
description="imsitu VSRL. Training, evaluation and prediction.")
parser.add_argument("--gpuid",
default=-1,
help="put GPU id > -1 in GPU mode",
type=int)
#parser.add_argument("--command", choices = ["train", "eval", "resume", 'predict'], required = True)
parser.add_argument('--resume_training',
action='store_true',
help='Resume training from the model [resume_model]')
parser.add_argument('--resume_model',
type=str,
default='',
help='The model we resume')
parser.add_argument('--pretrained_buatt_model',
type=str,
default='',
help='pretrained verb module')
parser.add_argument('--train_role',
action='store_true',
help='cnn fix, verb fix, role train from the scratch')
parser.add_argument(
'--use_pretrained_buatt',
action='store_true',
help='cnn fix, verb finetune, role train from the scratch')
parser.add_argument(
'--finetune_cnn',
action='store_true',
help='cnn finetune, verb finetune, role train from the scratch')
parser.add_argument('--output_dir',
type=str,
default='./trained_models',
help='Location to output the model')
parser.add_argument('--evaluate',
action='store_true',
help='Only use the testing mode')
parser.add_argument('--test',
action='store_true',
help='Only use the testing mode')
parser.add_argument('--dataset_folder',
type=str,
default='./imSitu',
help='Location of annotations')
parser.add_argument('--imgset_dir',
type=str,
default='./resized_256',
help='Location of original images')
parser.add_argument('--frcnn_feat_dir',
type=str,
help='Location of output from detectron')
parser.add_argument('--train_file',
default="train_new_2000_all.json",
type=str,
help='trainfile name')
parser.add_argument('--dev_file',
default="dev_new_2000_all.json",
type=str,
help='dev file name')
parser.add_argument('--test_file',
default="test_new_2000_all.json",
type=str,
help='test file name')
parser.add_argument('--model_saving_name',
type=str,
help='save name of the outpul model')
parser.add_argument('--epochs', type=int, default=500)
parser.add_argument('--num_hid', type=int, default=1024)
parser.add_argument(
'--model',
type=str,
default='baseline0grid_imsitu_verbiter_resnetfeatextract')
parser.add_argument('--output', type=str, default='saved_models/exp0')
parser.add_argument('--batch_size', type=int, default=64)
parser.add_argument('--seed', type=int, default=1111, help='random seed')
parser.add_argument('--num_iter_verb', type=int, default=2)
parser.add_argument('--num_iter_role', type=int, default=1)
parser.add_argument('--role_module',
type=str,
default='',
help='pretrained role module')
#verb special args
#todo: train role module separately with gt verbs
args = parser.parse_args()
clip_norm = 0.25
n_epoch = args.epochs
batch_size = args.batch_size
n_worker = 3
#dataset_folder = 'imSitu'
#imgset_folder = 'resized_256'
dataset_folder = args.dataset_folder
imgset_folder = args.imgset_dir
print('model spec :iter verb1')
train_set = json.load(open(dataset_folder + '/' + args.train_file))
imsitu_roleq = json.load(open("data/imsitu_questions_prev.json"))
verbq_dict_path = 'data/dictionary_imsitu_verbqagents.pkl'
verbq_dictionary = Dictionary.load_from_file(verbq_dict_path)
verbq_w_emb_path = 'data/glove6b_init_imsitu_verbqagents300d.npy'
role_dict_path = 'data/dictionary_imsitu_roleall.pkl'
role_dictionary = Dictionary.load_from_file(role_dict_path)
role_w_emb_path = 'data/glove6b_init_imsitu_roleall_300d.npy'
encoder = imsitu_encoder(train_set, imsitu_roleq, role_dictionary,
verbq_dictionary)
train_set = imsitu_loader_verb_buatt_iter4cnn(imgset_folder, train_set,
encoder, 'train',
encoder.train_transform)
#get role_model
print('loading role model')
role_constructor = 'build_%s' % 'baseline0grid_imsitu4verb'
role_model = getattr(base_model,
role_constructor)(train_set, args.num_hid,
encoder.get_num_labels(), encoder,
args.num_iter_role)
#utils_imsitu.load_net(args.role_module, [role_model])
#utils_imsitu.set_trainable(role_model, False)
print('role model loaded')
constructor = 'build_%s' % args.model
model = getattr(base_model, constructor)(train_set, args.num_hid,
encoder.get_num_verbs(), encoder,
role_model, args.num_iter_verb)
model.w_emb.init_embedding(verbq_w_emb_path)
train_loader = torch.utils.data.DataLoader(train_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
dev_set = json.load(open(dataset_folder + '/' + args.dev_file))
dev_set = imsitu_loader_verb_buatt_iter4cnn(imgset_folder, dev_set,
encoder, 'dev',
encoder.dev_transform)
dev_loader = torch.utils.data.DataLoader(dev_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
test_set = json.load(open(dataset_folder + '/' + args.test_file))
test_set = imsitu_loader_verb_buatt_iter4cnn(imgset_folder, test_set,
encoder, 'test',
encoder.dev_transform)
test_loader = torch.utils.data.DataLoader(test_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
if not os.path.exists(args.output_dir):
os.mkdir(args.output_dir)
torch.manual_seed(1234)
if args.gpuid >= 0:
#print('GPU enabled')
model.cuda()
torch.cuda.manual_seed(1234)
torch.backends.cudnn.deterministic = True
if args.resume_training:
print('Resume training from: {}'.format(args.resume_model))
args.train_all = True
if len(args.resume_model) == 0:
raise Exception('[pretrained module] not specified')
utils_imsitu.load_net(args.resume_model, [model])
model_name = 'resume_all'
utils_imsitu.set_trainable(model, False)
#optimizer = torch.optim.SGD(model.parameters(), lr=0.001, momentum=0.9)
if args.gpuid >= 0:
model.cuda()
extract_features(model, 'train', train_loader, args.gpuid,
len(train_loader) * batch_size)
extract_features(model, 'val', dev_loader, args.gpuid,
len(dev_loader) * batch_size)
extract_features(model, 'test', test_loader, args.gpuid,
len(test_loader) * batch_size)
def main():
import argparse
parser = argparse.ArgumentParser(
description="imsitu VSRL. Training, evaluation and prediction.")
parser.add_argument("--gpuid",
default=-1,
help="put GPU id > -1 in GPU mode",
type=int)
#parser.add_argument("--command", choices = ["train", "eval", "resume", 'predict'], required = True)
parser.add_argument('--resume_training',
action='store_true',
help='Resume training from the model [resume_model]')
parser.add_argument('--resume_model',
type=str,
default='',
help='The model we resume')
parser.add_argument('--pretrained_buatt_model',
type=str,
default='',
help='pretrained verb module')
parser.add_argument('--train_role',
action='store_true',
help='cnn fix, verb fix, role train from the scratch')
parser.add_argument(
'--use_pretrained_buatt',
action='store_true',
help='cnn fix, verb finetune, role train from the scratch')
parser.add_argument(
'--finetune_cnn',
action='store_true',
help='cnn finetune, verb finetune, role train from the scratch')
parser.add_argument('--output_dir',
type=str,
default='./trained_models',
help='Location to output the model')
parser.add_argument('--evaluate',
action='store_true',
help='Only use the testing mode')
parser.add_argument('--test',
action='store_true',
help='Only use the testing mode')
parser.add_argument('--dataset_folder',
type=str,
default='./imSitu',
help='Location of annotations')
parser.add_argument('--imgset_dir',
type=str,
default='./resized_256',
help='Location of original images')
parser.add_argument('--frcnn_feat_dir',
type=str,
help='Location of output from detectron')
parser.add_argument('--train_file',
default="train_new_2000_all.json",
type=str,
help='trainfile name')
parser.add_argument('--dev_file',
default="dev_new_2000_all.json",
type=str,
help='dev file name')
parser.add_argument('--test_file',
default="test_new_2000_all.json",
type=str,
help='test file name')
parser.add_argument('--model_saving_name',
type=str,
help='save name of the outpul model')
parser.add_argument('--epochs', type=int, default=500)
parser.add_argument('--num_hid', type=int, default=1024)
parser.add_argument('--model',
type=str,
default='baseline0grid_imsitu_agent')
parser.add_argument('--output', type=str, default='saved_models/exp0')
parser.add_argument('--batch_size', type=int, default=64)
parser.add_argument('--num_iter', type=int, default=1)
parser.add_argument('--seed', type=int, default=1111, help='random seed')
#todo: train role module separately with gt verbs
args = parser.parse_args()
clip_norm = 0.25
n_epoch = args.epochs
batch_size = args.batch_size
n_worker = 3
#dataset_folder = 'imSitu'
#imgset_folder = 'resized_256'
dataset_folder = args.dataset_folder
imgset_folder = args.imgset_dir
print('model spec :, top down att with role q ')
train_set = json.load(open(dataset_folder + '/' + args.train_file))
imsitu_roleq = json.load(open("data/imsitu_questions_prev.json"))
dict_path = 'data/dictionary_imsitu_roleall.pkl'
dictionary = Dictionary.load_from_file(dict_path)
w_emb_path = 'data/glove6b_init_imsitu_roleall_300d.npy'
encoder = imsitu_encoder(train_set, imsitu_roleq, dictionary)
train_set = imsitu_loader_roleq_buatt_place(imgset_folder, train_set,
encoder, dictionary, 'train',
encoder.train_transform)
constructor = 'build_%s' % args.model
model = getattr(base_model, constructor)(train_set, args.num_hid,
len(encoder.place_label_list),
encoder)
model.w_emb.init_embedding(w_emb_path)
#print('MODEL :', model)
train_loader = torch.utils.data.DataLoader(train_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
dev_set = json.load(open(dataset_folder + '/' + args.dev_file))
dev_set = imsitu_loader_roleq_buatt_place(imgset_folder, dev_set, encoder,
dictionary, 'val',
encoder.dev_transform)
dev_loader = torch.utils.data.DataLoader(dev_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
test_set = json.load(open(dataset_folder + '/' + args.test_file))
test_set = imsitu_loader_roleq_buatt_place(imgset_folder, test_set,
encoder, dictionary, 'test',
encoder.dev_transform)
test_loader = torch.utils.data.DataLoader(test_set,
batch_size=batch_size,
shuffle=True,
num_workers=n_worker)
if not os.path.exists(args.output_dir):
os.mkdir(args.output_dir)
torch.manual_seed(1234)
if args.gpuid >= 0:
#print('GPU enabled')
model.cuda()
torch.cuda.manual_seed(1234)
torch.backends.cudnn.deterministic = True
if args.use_pretrained_buatt:
print('Use pretrained from: {}'.format(args.pretrained_buatt_model))
if len(args.pretrained_buatt_model) == 0:
raise Exception('[pretrained buatt module] not specified')
#model_data = torch.load(args.pretrained_ban_model, map_location='cpu')
#model.load_state_dict(model_data.get('model_state', model_data))
utils_imsitu.load_net_ban(args.pretrained_buatt_model, [model],
['module'], ['w_emb', 'classifier'])
model_name = 'pre_trained_buatt'
elif args.resume_training:
print('Resume training from: {}'.format(args.resume_model))
args.train_all = True
if len(args.resume_model) == 0:
raise Exception('[pretrained module] not specified')
utils_imsitu.load_net(args.resume_model, [model])
optimizer_select = 0
model_name = 'resume_all'
else:
print('Training from the scratch.')
model_name = 'train_full'
utils_imsitu.set_trainable(model, True)
#utils_imsitu.set_trainable(model.classifier, True)
#utils_imsitu.set_trainable(model.w_emb, True)
#utils_imsitu.set_trainable(model.q_emb, True)
optimizer = torch.optim.Adamax([
{
'params': model.classifier.parameters()
},
{
'params': model.w_emb.parameters()
},
{
'params': model.q_emb.parameters(),
'lr': 5e-4
},
{
'params': model.v_att.parameters(),
'lr': 5e-5
},
{
'params': model.q_net.parameters(),
'lr': 5e-5
},
{
'params': model.v_net.parameters(),
'lr': 5e-5
},
],
lr=1e-3)
#utils_imsitu.set_trainable(model, True)
#optimizer = torch.optim.Adamax(model.parameters(), lr=1e-3)
#optimizer = torch.optim.Adam(model.parameters(), lr=lr, weight_decay=weight_decay)
#scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=lr_step, gamma=lr_gamma)
#gradient clipping, grad check
scheduler = torch.optim.lr_scheduler.ExponentialLR(optimizer, gamma=0.9)
if args.evaluate:
top1, top5, val_loss = eval(model,
dev_loader,
encoder,
args.gpuid,
write_to_file=True)
top1_avg = top1.get_average_results_nouns()
top5_avg = top5.get_average_results_nouns()
avg_score = top1_avg["verb"] + top1_avg["value"] + top1_avg["value-all"] + top5_avg["verb"] + \
top5_avg["value"] + top5_avg["value-all"] + top5_avg["value*"] + top5_avg["value-all*"]
avg_score /= 8
print('Dev average :{:.2f} {} {}'.format(
avg_score * 100, utils_imsitu.format_dict(top1_avg, '{:.2f}',
'1-'),
utils_imsitu.format_dict(top5_avg, '{:.2f}', '5-')))
#write results to csv file
role_dict = top1.role_dict
fail_val_all = top1.value_all_dict
pass_val_dict = top1.vall_all_correct
with open('role_pred_data.json', 'w') as fp:
json.dump(role_dict, fp, indent=4)
with open('fail_val_all.json', 'w') as fp:
json.dump(fail_val_all, fp, indent=4)
with open('pass_val_all.json', 'w') as fp:
json.dump(pass_val_dict, fp, indent=4)
print('Writing predictions to file completed !')
elif args.test:
top1, top5, val_loss = eval(model,
test_loader,
encoder,
args.gpuid,
write_to_file=True)
top1_avg = top1.get_average_results_nouns()
top5_avg = top5.get_average_results_nouns()
avg_score = top1_avg["verb"] + top1_avg["value"] + top1_avg["value-all"] + top5_avg["verb"] + \
top5_avg["value"] + top5_avg["value-all"] + top5_avg["value*"] + top5_avg["value-all*"]
avg_score /= 8
print('Test average :{:.2f} {} {}'.format(
avg_score * 100, utils_imsitu.format_dict(top1_avg, '{:.2f}',
'1-'),
utils_imsitu.format_dict(top5_avg, '{:.2f}', '5-')))
else:
print('Model training started!')
train(model, train_loader, dev_loader, None, optimizer, scheduler,
n_epoch, args.output_dir, encoder, args.gpuid, clip_norm, None,
model_name, args.model_saving_name, args)
args = parse_args()
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
torch.backends.cudnn.benchmark = True
dictionary = Dictionary.load_from_file('data/dictionary.pkl')
image_filenames = json.load(open("data/coco_img_file_mapping.json"))
train_dset = VQAGridDataset(args.imgset_dir + '/train2014',
image_filenames, 'train', dictionary)
eval_dset = VQAGridDataset(args.imgset_dir + '/val2014', image_filenames,
'val', dictionary)
batch_size = args.batch_size
constructor = 'build_%s' % args.model
model = getattr(base_model, constructor)(train_dset, args.num_hid).cuda()
utils_imsitu.set_trainable(model.conv_net, False)
model.w_emb.init_embedding('data/glove6b_init_300d.npy')
model = nn.DataParallel(model).cuda()
train_loader = DataLoader(train_dset,
batch_size,
shuffle=True,
num_workers=3)
eval_loader = DataLoader(eval_dset,
batch_size,
shuffle=True,
num_workers=3)
train(model, train_loader, eval_loader, args.epochs, args.output)
