Dataset Creation

#2
by NikitaKras - opened

Hi

May I ask how this dataset was created? Primarily, I am curious how the summarization was produced: either manually or auto-generated with GPT/Cloude/other LLM.

Thanks,
Nikita Krasnytskyi

Hi @NikitaKras
The source code that daily updates the dataset is referenced in the readme.
There's no summarization applied at the moment.
Data is fetched from the source and stored in the dataset as-is. The "Topic" field is just a header tag value from the source HTML.
If you have some inspiration and time, feel free to propose changes that introduce summarization.

I would be happy to add summarization if I manage to fine-tune the model to summarize Ukrainian text well enough. I will update you in case I succeed.

Sign up or log in to comment