README.md

metadata

license: mit
task_categories:
  - text-classification
tags:
  - text-classification
  - politeness
  - nlp
  - active-learning
  - fine-tuning
  - data-centric-ai

Stanford Politeness Dataset

This dataset contains politeness classification data based on the Stanford Politeness Corpus for active learning and fine-tuning tasks.

Dataset Description

The dataset is organized into two main directories:

Active Learning

• X_labeled_full.csv - Labeled examples
• X_unlabeled.csv - Unlabeled examples for active learning
• extra_annotations.npy - Additional annotation data
• test.csv - Test set

Fine-tuning

• train.csv - Training set
• train_fixed.csv - Fixed training set
• train_full.csv - Full training set
• test.csv - Test set

Usage

import pandas as pd
from huggingface_hub import hf_hub_download

# Download a specific file
file_path = hf_hub_download(
    repo_id="Cleanlab/stanford-politeness",
    filename="fine-tuning/train.csv",
    repo_type="dataset"
)
df = pd.read_csv(file_path)
print(df.head())

License

MIT License