Ecosyste.ms: OpenCollective

An open API service for software projects hosted on Open Collective.

github.com/mozilla/translations

The code, training pipeline, and models that power Firefox Translations
https://github.com/mozilla/translations

bicleaner-ai crashes with out of GPU memory, when running with more than one jobs per node

XapaJIaMnu opened this issue about 3 years ago

deescape-special-chars.perl should be part of the clean step

XapaJIaMnu opened this issue about 3 years ago

bicleaner should be optional for some corpora

XapaJIaMnu opened this issue about 3 years ago

spm vocabulary training could fail on high res languages

XapaJIaMnu opened this issue about 3 years ago

find-corpus.py is outdated?

XapaJIaMnu opened this issue about 3 years ago

Change sharding mode in training to improve speed

XapaJIaMnu opened this issue about 3 years ago

Version of marian used for training

XapaJIaMnu opened this issue about 3 years ago

Quality improvements

eu9ene opened this pull request about 3 years ago

Add chrF evaluation metric

eu9ene opened this issue about 3 years ago

Evaluate ensemble of teachers

eu9ene opened this issue about 3 years ago

Handle soft hyphens with custom normalization tables

eu9ene opened this issue about 3 years ago

Remove punctuation normalizaiton

eu9ene opened this issue about 3 years ago

Snakemake integration

eu9ene opened this pull request about 3 years ago

Fine tune teacher on parallel data only

eu9ene opened this issue about 3 years ago

Add dataset specific bicleaner thresholds

eu9ene opened this issue about 3 years ago

Out of memory on shuffling huge datasets