Credits of the datasets that the landing page shows and our tests use

The part How it works of the landing page shows rows of these three datasets, and our tests use them. We show each row as the
dataset gives it. We give each field and each label a name in plain words.

1. BANKING77

Creators: Casanueva and others, PolyAI, 2020 (Efficient Intent Detection with Dual Sentence Encoders).
Source: https://huggingface.co/datasets/PolyAI/banking77
Licence: Creative Commons Attribution 4.0 International (CC BY 4.0), https://creativecommons.org/licenses/by/4.0/
Changes: The page shows four customer messages and their labels in plain words. Our model learned from the messages and the
labels of the dataset, and our tests use messages that it did not learn from.

2. CORD v2

Creators: Park and others, NAVER Clova AI, 2019 (CORD: A Consolidated Receipt Dataset for Post-OCR Parsing).
Source: https://huggingface.co/datasets/naver-clova-ix/cord-v2, revision 7f0115a4b758a71d6473b8d085751692da2fef98
Licence: Creative Commons Attribution 4.0 International (CC BY 4.0), https://creativecommons.org/licenses/by/4.0/
Changes: We use the text lines of each receipt and its summary fields, not the images. The page shows the lines of four receipts
and their fields in plain words.

3. gretel-pii-masking-en-v1

Creator: Gretel.ai.
Source: https://huggingface.co/datasets/gretelai/gretel-pii-masking-en-v1, revision e06eb1499ca8d54470f085021cd8e54f9efac7fd
Licence: Apache License, Version 2.0, http://www.apache.org/licenses/LICENSE-2.0. The full text of this licence is in
/licences/qwen3-4b-instruct-2507-LICENSE.txt.
Changes: We use the documents and their marked personal data. The page shows four synthetic documents and the personal data in
them, with each type in plain words.