if args.load_check_point:
if os.path.isfile(args.load_check_point):
print("loading checkpoint: '{}'".format(args.load_check_point))
checkpoint_file = torch.load(args.load_check_point)
args.start_epoch = checkpoint_file['epoch']
f_map = checkpoint_file['f_map']
l_map = checkpoint_file['l_map']
train_features, train_labels = utils.read_corpus(lines)
else:
print("no checkpoint found at: '{}'".format(args.load_check_point))
else:
print('constructing coding table')
# converting format
train_features, train_labels, f_map, l_map = utils.generate_corpus(
lines, if_shrink_feature=True, thresholds=0)
f_set = {v for v in f_map}
f_map = utils.shrink_features(f_map, train_features, args.mini_count)
dt_f_set = functools.reduce(lambda x, y: x | y,
map(lambda t: set(t), dev_features), f_set)
if not args.rand_embedding:
print("feature size: '{}'".format(len(f_map)))
print('loading embedding')
if args.fine_tune: # which means does not do fine-tune
f_map = {'<eof>': 0}
f_map, embedding_tensor, in_doc_words = utils.load_embedding_wlm(
args.emb_file, ' ', f_map, dt_f_set, args.caseless, args.unk,
args.embedding_dim)
checkpoint_file = torch.load(args.load_check_point)
args.start_epoch = checkpoint_file['epoch']
f_map = checkpoint_file['f_map']
l_map = checkpoint_file['l_map']
lexicon_f_map = checkpoint_file['lexicon_f_map']
train_features, train_labels = utils.read_corpus(lines)
lexicon_train_features, lexicon_feature_map = utils.read_corpus_lexicon(
lexicon_train_lines, train_features, lexicon_feature_map)
else:
print("no checkpoint found at: '{}'".format(args.load_check_point))
else:
print('constructing coding table')
# converting format
train_features, train_labels, f_map, l_map, train_bichar_features, bichar_f_map = utils.generate_corpus(
lines, if_shrink_feature=False, thresholds=0)
f_set = {v for v in f_map}
f_map = utils.shrink_features(f_map, train_features, args.mini_count)
bichar_f_set = {v for v in bichar_f_map}
bichar_f_map = utils.shrink_features(bichar_f_map,
train_bichar_features,
args.mini_count)
lexicon_train_features, lexicon_f_map = utils.generous_corpus_lexicon(
lexicon_train_lines, train_features, lexicon_feature_map)
lexicon_f_set = {v for v in lexicon_f_map}
dt_f_set = functools.reduce(lambda x, y: x | y,
map(lambda t: set(t), dev_features), f_set)
f_map = checkpoint_file['f_map']
l_map = checkpoint_file['l_map']
a_map = checkpoint_file['a_map']
ner_map = checkpoint_file['ner_map']
char_map = checkpoint_file['char_map']
singleton = checkpoint_file['singleton']
train_features, train_labels, train_actions, word_count = utils.read_corpus_ner(
lines, word_count)
else:
print("no checkpoint found at: '{}'".format(args.load_check_point))
else:
print('constructing coding table')
train_features, train_labels, train_actions, f_map, l_map, a_map, ner_map, singleton, char_map = utils.generate_corpus(
lines,
word_count,
args.spelling,
if_shrink_feature=True,
thresholds=0)
f_set = {v for v in f_map}
dt_f_set = functools.reduce(
lambda x, y: x | y, map(lambda t: set(t), dev_features),
f_set) # Add word in dev and in test into feature_map
dt_f_set = functools.reduce(lambda x, y: x | y,
map(lambda t: set(t), test_features),
dt_f_set)
dt_f_set = functools.reduce(lambda x, y: x | y,
map(lambda t: set(t), train_features),
dt_f_set)
if not args.rand_embedding:
if args.load_check_point:
if os.path.isfile(args.load_check_point):
print("loading checkpoint: '{}'".format(args.load_check_point))
checkpoint_file = torch.load(args.load_check_point)
args.start_epoch = checkpoint_file['epoch']
f_map = checkpoint_file['f_map']
l_map = checkpoint_file['l_map']
train_features, train_labels = utils.read_corpus(lines)
else:
print("no checkpoint found at: '{}'".format(args.load_check_point))
else:
print('constructing coding table')
# converting format
train_features, train_labels, f_map, l_map = utils.generate_corpus(lines, if_shrink_feature=True, thresholds=0)
f_set = {v for v in f_map}
f_map = utils.shrink_features(f_map, train_features, args.mini_count)
dt_f_set = functools.reduce(lambda x, y: x | y, map(lambda t: set(t), dev_features), f_set)
dt_f_set = functools.reduce(lambda x, y: x | y, map(lambda t: set(t), test_features), dt_f_set)
if not args.rand_embedding:
print("feature size: '{}'".format(len(f_map)))
print('loading embedding')
if args.fine_tune: # which means does not do fine-tune
f_map = {'<eof>': 0}
f_map, embedding_tensor, in_doc_words = utils.load_embedding_wlm(args.emb_file, ' ', f_map, dt_f_set,args.caseless,args.unk, args.embedding_dim, shrink_to_corpus=args.shrink_embedding)
print("embedding size: '{}'".format(len(f_map)))