Skip to content
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

ensure transforms are ordered as declared #9

Merged
merged 1 commit into from
Sep 20, 2023
Merged

ensure transforms are ordered as declared #9

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
15 changes: 12 additions & 3 deletions onmt/inputters/dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -165,17 +165,26 @@ def get_corpus(opts, task, src_vocab: Vocab, tgt_vocab: Vocab, is_train: bool =
"""build an iterable Dataset object"""
# get transform classes to infer special tokens
# FIXME ensure TQM properly initializes transform with global if necessary
vocabs = {'src': src_vocab, 'tgt': tgt_vocab}
corpus_opts = opts.data[task.corpus_id]
transforms_cls = get_transforms_cls(corpus_opts.get('transforms', opts.transforms))
transforms_to_apply = corpus_opts.get('transforms', None)
transforms_to_apply = transforms_to_apply or opts.get('transforms', None)
transforms_to_apply = transforms_to_apply or []
transforms_cls = make_transforms(
opts,
get_transforms_cls(transforms_to_apply),
vocabs,
task=task,
)
transforms_to_apply = [transforms_cls[trf_name] for trf_name in transforms_to_apply]

vocabs = {'src': src_vocab, 'tgt': tgt_vocab}
# build Dataset proper
dataset = ParallelCorpus(
corpus_opts["path_src"] if is_train else corpus_opts["path_valid_src"],
corpus_opts["path_tgt"] if is_train else corpus_opts["path_valid_tgt"],
src_vocab,
tgt_vocab,
TransformPipe(opts, make_transforms(opts, transforms_cls, vocabs, task=task).values()),
TransformPipe(opts, transforms_to_apply),
stride=corpus_opts.get('stride', None),
offset=corpus_opts.get('offset', None),
is_train=is_train,
Expand Down