# Training data attribution

Loom Router 1 was trained on utterances from two openly licensed corpora. Both
licences require attribution; this file satisfies that requirement and must be
kept with any redistribution.

## MASSIVE — CC BY 4.0
Amazon. "MASSIVE: A 1M-Example Multilingual Natural Language Understanding
Dataset with 51 Typologically-Diverse Languages."
https://github.com/alexa/massive — licensed CC BY 4.0.
MASSIVE is derived from SLURP, also licensed CC BY 4.0.

## CLINC150 — CC BY 3.0
Larson et al. "An Evaluation Dataset for Intent Classification and Out-of-Scope
Prediction." https://github.com/clinc/oos-eval — licensed CC BY 3.0.

Only the English portions were used. Intent labels were remapped onto the 17
Loom routes; no utterance text was altered except for surface augmentation
(casing, punctuation, filler) applied to training copies only.