license: apache-2.0 | |
language: | |
- es | |
- gn | |
metrics: | |
- chrf | |
tags: | |
- marianNMT | |
- translation | |
- grammar | |
*Description* | |
This model was created using the same methodology presented in [Grammar-based Data Augmentation for Low-Resource Languages: The Case of Guarani-Spanish Neural Machine Translation](https://aclanthology.org/2024.naacl-long.354/). | |
*Instructions* | |
In order to run it, it is needed to use the C++ software [MarianNMT](https://marian-nmt.github.io/) with a command similar to this: | |
``` | |
/marian/build/marian-decoder -m ./s2s_adjusted_gn_es_s2s.npz -v ./pretrain_test_vocab.gnV6000_6000.spm ./pretrain_test_vocab.esV6000_6000.spm -i ./input.gn --cpu-threads 12 > ./output.es | |
``` | |
*Benchmarks* | |
- 48.98 ChrF in [Jojajovai](https://aclanthology.org/2022.lrec-1.226/) |