NVIDIA Releases Open Multilingual Synthetic Personas Dataset For Belgium · Digg

NVIDIA Releases Open Multilingual Synthetic Personas Dataset For Belgium · Digg

Posts from X

Most Activity

VIEWS419BOOKMARKS1LIKES8REPLIES2

Alexander Doria@Dorialexander

Thanks to @pieterdelobelle, we managed to model uniquely complex European dynamics at a smaller scale. The personas had to reflect the demographics of three federated regions, three language communities using a combination of data from Statbel and regional authorities.

Alexander Doria@Dorialexander

And new data release: in partnership with @nvidia we publish at the second open synthetic dataset for personas in Europe, Nemotron-Personas-Belgium and the first fully multilingual (in French, Dutch and German).

2h41981

Alexander Doria@Dorialexander

@pieterdelobelle @nvidia Full dataset on @huggingface: https://huggingface.co/datasets/nvidia/Nemotron-Personas-Belgium Blogpost announcement: https://pleias.ai/blog/nvidia-nemotron-personas-belgium

Alexander Doria@Dorialexander

@pieterdelobelle For the first time we generated the synthetic lives of each personas on European public compute at Jean Zay. Multilingual alignment was a significant challenge and with @nvidia collaboration we strived to build a reusable blueprint for other linguistically rich countries

2h32850

Alexander Doria@Dorialexander

@pieterdelobelle @nvidia @huggingface And more details from @pieterdelobelle

Pieter Delobelle@pieterdelobelle

Today we release Nemotron-Personas-Belgium in collaboration with @NVIDIAAI , a synthetic dataset of 4x300k Belgian personas in Dutch, French, German and English.

The dataset is modeled on the @Statbel_en census, modeling age, occupations, household, income, names etc.

2h34221

Alexander Doria@Dorialexander

@pieterdelobelle For the first time we generated the synthetic lives of each personas on European public compute at Jean Zay. Multilingual alignment was a significant challenge and with @nvidia collaboration we strived to build a reusable blueprint for other linguistically rich countries

2h1613

Pieter Delobelle@pieterdelobelle

@Dorialexander proud to have have worked on this 🫡

2h1

Sophie Xhonneux@SophieXhon11060

@Dorialexander @pieterdelobelle @nvidia @huggingface Belgium represent!

2h1