Skip to content

Dataset · SetFit

5-class sentiment annotated texts (SST-5)

SetFit/sst5

A classic text corpus ranging from 10k to 100k sentences for fine-grained sentiment analysis.

What's inside

The dataset is a version of the Stanford Sentiment Treebank divided into five sentiment categories, ranging from very negative to very positive. Unlike basic versions, the training split here is annotated at the whole-sentence level rather than individual phrases. The split division is taken from the AcademiaSinicaNLPLab repository, with a total volume ranging from 10K to 100K entries.

What you can do with it

  • Training and fine-tuning text classifiers for granular review analysis
  • Comparing language model accuracy on a standard sentiment analysis benchmark
Size
10K–100K rows
Downloads this month
20.4K
Likes
24

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("SetFit/sst5")