Adding an end of sentence symbol to the source side. #392
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
fhieber
reviewed
May 15, 2018
sockeye/data_io.py
Outdated
| yield sequence | ||
|
|
||
|
|
||
| def create_sequence_readers(sources, target, vocab_sources, vocab_target): |
There was a problem hiding this comment.
You could add type annotations.
sockeye/inference.py
Outdated
| @@ -423,6 +427,11 @@ def load_models(context: mx.context.Context, | |||
| utils.check_condition(vocab.are_identical(*[source_vocabs[i][fi] for i in range(len(source_vocabs))]), | |||
| "Source vocabulary ids do not match. Factor %d" % fi) | |||
|
|
|||
| source_with_eos = models[0].source_with_eos | |||
| utils.check_condition(all(source_with_eos == m.source_with_eos for m in models), | |||
| "All models must match either take an additional EOS symbol on the source side or not. " | |||
There was a problem hiding this comment.
grammar, maybe: "All models must agree on using source-side EOS symbols or not"
There was a problem hiding this comment.
sounds good, changed.
sockeye/inference.py
Outdated
| @@ -591,6 +600,13 @@ def chunks(self, chunk_size: int) -> Generator['TranslatorInput', None, None]: | |||
| factors=factors, | |||
| chunk_id=chunk_id) | |||
|
|
|||
| def with_eos(self): | |||
sockeye/train.py
Outdated
| @@ -258,12 +260,14 @@ def create_data_iters_and_vocabs(args: argparse.Namespace, | |||
| Create the data iterators and the vocabularies. | |||
|
|
|||
| :param args: Arguments as returned by argparse. | |||
| :param max_seq_len_source: Source maximum sequence length. | |||
| :param max_seq_len_target: Target maximum sequence length. | |||
| :param shared_vocab: Whether to create a shared vocabulary. | |||
There was a problem hiding this comment.
double docstring for shared_vocab
|
1 unit test is failing (test_arguments.py) |
|
true, should be fixed now. |
|
they seem to work now :) |
|
Feel free to merge when travis or system tests are ready. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
With this change we add an additional EOS symbol on the source side. From the CLI/user perspective the
--max-seq-lenis the maximum length of the raw tokens without special symbols. Internally lengths are now based on the lengths including the special symbols like EOS. The change is backwards compatible in that the inference logic for existing models did not change.System tests are currently running.
Pull Request Checklist
until you can check this box.
pytest)pytest test/system)./style-check.sh)sockeye/__init__.py. Major version bump if this is a backwards incompatible change.By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.