xendalm/

taiga-sum

This dataset contains synthetic summaries generated by the gemini-2.0-flash model. Source texts come from the nplus1 and fontanka segments of the Taiga Corpus. All texts were cleaned of artifacts, fontanka segment was also filtered to keep only articles between 300 and 600 words long.

	Splits Statistics




	


	nplus1




	

input_words target_words compression_words

mean 224.12 64.43 3.383

std 106.4315.66 1.140

25%… See the full description on the dataset page: https://huggingface.co/datasets/xendalm/taiga-sum.