Skip to content

Update the order of the datasets in the BERT example #1040

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Oct 19, 2020
Merged

Update the order of the datasets in the BERT example #1040

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6b9c0e1
adjust the order of the datasets
Oct 14, 2020
9f90a27
Merge branch 'master' into update_BERT
Oct 14, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion examples/BERT/data.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@
# Set up dataset for book corpus
###################################################################
def BookCorpus(vocab, tokenizer=get_tokenizer("basic_english"),
data_select=('train', 'test', 'valid'), removed_tokens=[],
data_select=('train', 'valid', 'test'), removed_tokens=[],
min_sentence_len=None):

if isinstance(data_select, str):
Expand Down
6 changes: 3 additions & 3 deletions examples/BERT/mlm_task.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -123,15 +123,15 @@ def run_main(args, rank=None):
try:
vocab = torch.load(args.save_vocab)
except:
train_dataset, test_dataset, valid_dataset = WLMDataset()
train_dataset, valid_dataset, test_dataset = WLMDataset()
old_vocab = train_dataset.vocab
vocab = torchtext.vocab.Vocab(counter=old_vocab.freqs,
specials=['<unk>', '<pad>', '<MASK>'])
with open(args.save_vocab, 'wb') as f:
torch.save(vocab, f)

if args.dataset == 'WikiText103' or args.dataset == 'WikiText2':
train_dataset, test_dataset, valid_dataset = WLMDataset(vocab=vocab)
train_dataset, valid_dataset, test_dataset = WLMDataset(vocab=vocab)
elif args.dataset == 'WMTNewsCrawl':
from torchtext.experimental.datasets import WikiText2
test_dataset, valid_dataset = WikiText2(vocab=vocab, data_select=('test', 'valid'))
Expand All @@ -150,7 +150,7 @@ def run_main(args, rank=None):
valid_dataset = LanguageModelingDataset(val_data, vocab)
test_dataset = LanguageModelingDataset(test_data, vocab)
elif args.dataset == 'BookCorpus':
train_dataset, test_dataset, valid_dataset = BookCorpus(vocab)
train_dataset, valid_dataset, test_dataset = BookCorpus(vocab)

train_data = process_raw_data(train_dataset.data, args)
if rank is not None:
Expand Down
2 changes: 1 addition & 1 deletion examples/BERT/ns_task.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,7 +141,7 @@ def run_main(args, rank=None):
single_line=False)
elif args.dataset == 'BookCorpus':
from data import BookCorpus
train_dataset, test_dataset, valid_dataset = BookCorpus(vocab, min_sentence_len=60)
train_dataset, valid_dataset, test_dataset = BookCorpus(vocab, min_sentence_len=60)

if rank is not None:
chunk_len = len(train_dataset.data) // args.world_size
Expand Down