How de9uch1/fairseq-tutorial works
This repository is a Fairseq tutorial, not the Fairseq implementation: its README describes the toolkit and documents translation training and generation, while the tree contains a small translation example. The documented user journey is to prepare translation data, train a model, generate translations, and evaluate them. The example directory also contains configuration and text-processing assets, but their internal use is not established by the available excerpts; those relationships are therefore omitted.
Main entry points: README.rst Fairseq tutorial · translation/README.rst Example guide
Components 9
Tutorial guide
- Fairseq tutorial
README.rst
Translation example
- Example guide
examples/translation/README.rst - Prepare data
examples/translation/prepare-aspec.sh - Transformer config
examples/translation/transformer.yaml - Detokenize output
examples/translation/detokenizer.sh - Evaluate translations
examples/translation/evaluate-aspec.sh - Text conversion
examples/translation/z2h-utf8.pl
Other components
- Researcher
- Fairseq toolkit
Connections 3
- Researcher → Fairseq tutorial: consults
README.rst - Fairseq tutorial → Fairseq toolkit: documents workflow
README.rst - Researcher → Example guide: opens example