ID | Benchmark (bleu) | Output | OPUS-MT | bleu | external | bleu | Diff |
---|---|---|---|---|---|---|---|
0 | flores101-devtest | compare | eng-lit/opus...2022-02-25 | 28 | facebook/m2m100_1.2B | 25.5 | 2.5 |
1 | flores200-devtest | compare | deu+eng+fra+...2024-05-30 | 28.2 | facebook/m2m100_1.2B | 25.5 | 2.7 |
2 | newstest2019 | compare | deu+eng+fra+...2024-05-30 | 18.9 | facebook/m2m100_1.2B | 17.4 | 1.5 |
3 | ntrex128 | compare | deu+eng+fra+...2024-05-30 | 21.9 | facebook/nll...illed-1.3B | 17.4 | 4.5 |
4 | tatoeba-test-v2020-07-28 | compare | deu+eng+fra+...2024-05-30 | 39.7 | facebook/nllb-200-3.3B | 37.6 | 2.1 |
5 | tatoeba-test-v2021-03-30 | compare | en-lt/opus-2019-12-04 | 39.6 | facebook/nllb-200-3.3B | 37.9 | 1.7 |
6 | tatoeba-test-v2021-08-07 | compare | deu+eng+fra+...2024-05-30 | 39.7 | facebook/nll...illed-1.3B | 37.2 | 2.5 |
average | 30.9 | 28.4 | 2.5 |