Datasets:

CZLC
/

sumeczech_downsampled

Tasks:

Modalities:

Formats:

Languages:

Size:

Libraries:

License:

Dataset card Data Studio Files Files and versions

mdocekal commited on May 17, 2024

Commit

c3c15dd

·

verified ·

1 Parent(s): 8821383

Update README.md

Files changed (1) hide show

README.md +3 -1

README.md CHANGED Viewed

@@ -10,7 +10,9 @@ pretty_name: SumeCzech
 <!-- Provide a quick summary of the dataset. -->
- SumeCzech is a 1-million-document dataset of Czech news, each consisting of:
  - headline;
  - abstract (visually distinguished first paragraph);
  - rest of the text.

 <!-- Provide a quick summary of the dataset. -->
+This is a downsampled version of the SumeCzech summarization dataset. The train, dev, and test sets each have 1500 samples.
+Original SumeCzech is a 1-million-document dataset of Czech news, each consisting of:
  - headline;
  - abstract (visually distinguished first paragraph);
  - rest of the text.