Skip to content
This repository was archived by the owner on Jul 7, 2023. It is now read-only.

Conversation

@martinpopel
Copy link
Contributor

  • simplify wmt.py:
    train_generator is now a method (not a property returning a function)
    and WMTProblem subclasses implement it directly (not via one-purpose functions)
  • add WMT English-Czech translation problems
  • introduce vocab_name property, so e.g. tokens.vocab-en-cs.32768 is used instead of tokens.vocab.32768 if only English+Czech datasets were used

`train_generator` is now a method (not a property returning a function)
and WMTProblem subclasses implement it directly (not via one-purpose functions)
@property
def train_generator(self):
"""Generator; takes tmp_dir, is_training, possibly targeted_vocab_size."""
def train_generator(self, tmp_dir, is_training):
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Unluckily these arguments now include data_dir, this will need updating. Also, they are different for character models, but it could be just an empty arg (_). Thanks for making this nicer!

@martinpopel
Copy link
Contributor Author

New attempt: #181

@martinpopel martinpopel deleted the add_czech branch July 25, 2017 19:37
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants