Skip to content

Dataset · knkarthick

DialogSum: dialogue summarization

knkarthick/dialogsum

13,460 English dialogues with summaries for training and evaluating models.

What's inside

A dataset for dialogue summarization in English. It contains 13,460 dialogues plus 100 held-out examples for topic generation. The texts are intended for summarization and text-generation tasks.

What you can do with it

  • Fine-tune a dialogue summarization model using SFT
  • Evaluate the quality of brief dialogue summaries
  • Create a LoRA for conversational speech summarization
Tasks
Summarization, Text generation
Languages
English
Size
10K–100K rows
Downloads this month
9.2K
Likes
259

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("knkarthick/dialogsum")