Constrained Data

[Samuel Larkin]

Hi,

  I would like to clarify what can be used to train a model.  Are we strictly limited to data from:

1. https://github.com/TUM-NLP/llms-limited-resources2025
2. https://github.com/mariondimarco/WMT22_UnsupVeryLowResMT_Data/

Are we allowed, for example, to use other resources from different languages (i.e., not DSB/HSB)?

Can I used another Question and Answer dataset that is not dsb/hsb/de?

Can I used, for example, de-pl bilingual text?

Thanks

[Daryna Dementieva]

Dear all,

Yes, we allow additional data usage. For fairness and reproducibility, the resources should, however, be publicly available. 

Best,

Organizers

вторник, 27 мая 2025 г. в 17:43:43 UTC+2, samuel...@gmail.com: