| ID | Benchmark (bleu) | Output | OPUS-MT | bleu | external | bleu | Diff |
|---|---|---|---|---|---|---|---|
| 0 | flores101-devtest | compare | eng-heb/opus-2021-02-18 | 28.4 | sshleifer/opus-mt-en-he | 28.5 | -0.1 |
| 1 | flores200-devtest | compare | deu+eng+fra+...2024-05-30 | 32.3 | sshleifer/opus-mt-en-he | 28.7 | 3.6 |
| 2 | ntrex128 | compare | deu+eng+fra+...2024-05-30 | 23.9 | facebook/nll...illed-1.3B | 21.1 | 2.8 |
| 3 | tatoeba-test-v2020-07-28 | compare | en-he/opus-2019-12-18 | 41.3 | Rahulrr/lang...odel_en_he | 37.8 | 3.5 |
| 4 | tatoeba-test-v2021-03-30 | compare | en-he/opus-2019-12-18 | 41.5 | Rahulrr/lang...odel_en_he | 38.5 | 3.0 |
| 5 | tatoeba-test-v2021-08-07 | compare | en-he/opus-2019-12-18 | 41.6 | Rahulrr/lang...odel_en_he | 38.6 | 3.0 |
| average | 34.8 | 32.2 | 2.6 |