| ID | Benchmark (bleu) | Output | OPUS-MT | bleu | external | bleu | Diff |
|---|---|---|---|---|---|---|---|
| 0 | flores101-devtest | compare | eng-pol/opus...2021-04-14 | 19.9 | facebook/nll...illed-1.3B | 19.6 | 0.3 |
| 1 | flores200-devtest | compare | deu+eng+fra+...2024-05-30 | 21.8 | facebook/nll...illed-1.3B | 19.6 | 2.2 |
| 2 | newstest2020 | compare | deu+eng+fra+...2024-05-30 | 24.9 | facebook/m2m100_1.2B | 23.3 | 1.6 |
| 3 | ntrex128 | compare | deu+eng+fra+...2024-05-30 | 25.8 | facebook/nll...illed-1.3B | 23.8 | 2.0 |
| 4 | tatoeba-test-v2020-07-28 | compare | en-pl/opus-2019-12-04 | 47.7 | gsarti/opus-mt-tc-en-pl | 47.5 | 0.2 |
| 5 | tatoeba-test-v2021-03-30 | compare | en-pl/opus-2019-12-04 | 47.8 | gsarti/opus-mt-tc-en-pl | 47.5 | 0.3 |
| 6 | tatoeba-test-v2021-08-07 | compare | en-pl/opus-2019-12-04 | 47.9 | gsarti/opus-mt-tc-en-pl | 47.6 | 0.3 |
| average | 33.7 | 32.7 | 1.0 |