The question behind the work
A translation experiment using Command R+ and FLORES-200 parallel evaluation data.
NLP & RESEARCH
Translation · Command R+
Translation and reference-based BLEU evaluation.
THE PROJECT
Translation · Command R+
Source and context from my portfolio
How it comes together
The experiment connects translation generation with reference-based evaluation. The model, sampling and language directions are defined in the notebook; related variants use other translation approaches.
What to keep in mind
BLEU measures overlap with references. It is useful to inspect it alongside the actual translations and language-specific context.