Datasets:
metadata
license: apache-2.0
task_categories:
- translation
language:
- lo
- vi
- zh
These 2 Datasets are from VLSP Evaluation Campaigns for Machine Translation Shared Task
1) VLSP 2022 Dataset
Task Description:
- Machine Translation Shared Task:
Chinese-Vietnamese Machine Translations (Chinese ↔ Vietnamese): The participants need to handle a scenario with limited data. Furthermore, since Chinese and Vietnamese can be considered similar languages to some degree (e.g., many Sino-Vietnamese words have a 1-to-1 mapping in meaning with their Chinese counterparts), participants could employ unique methods designed for similar language pairs in this task.
Bilingual Training and Evaluation Corpora Statistics:
| Task | Dataset | Sentences | Tokens (Vi) | Tokens (Zh) |
|---|---|---|---|---|
| Vi ↔ Zh | Train | 300,348 | 43,762 | 141,879 |
| Dev | 1,000 | 2,545 | 3,796 | |
| Test | 1,000 | 2,454 | 4,078 |
Link: VLSP 2022 Evaluation Campaign
2) VLSP 2023 Dataset
Task Description:
- Machine Translation Shared Task:
The 2023 shared task introduced two sub-tasks on Lao-Vietnamese Machine Translations (Lao ↔ Vietnamese). As in previous years, participants dealt with a scenario of limited data. Lao and Vietnamese share similarities (e.g., many Vietnamese words have a 1-to-1 mapping in meaning with their Lao counterparts), enabling participants to apply innovative methods for similar language pairs in this task.
Bilingual Training and Evaluation Corpora Statistics:
| Task | Dataset | Sentences | Tokens (Vi) | Tokens (Lo) |
|---|---|---|---|---|
| Vi ↔ Lo | Train | 100,000 | 2,850,755 | 4,070,248 |
| Dev | 1,000 | 52,821 | 72,789 | |
| Test | 1,000 | 32,386 | 49,175 |