Datasets
Open data, free to use.
Eval suites, training corpora, and standalone collections. Some are tied to our models. Many are not — they're just useful data we wanted to put in the open.
ChindaMT CoreEval
ReleasedTranslation eval
MT
Featured
400 paired Thai-English translation prompts across 5 deployment-grade domains, with plain and rules-following variants.
400 paired prompts JSONL text Thai English
Apr 28, 2026 View on HF
ChindaMT BroadEval
ReleasedTranslation eval
MT
400 paired Thai-English translation prompts across 10 domains — a cross-domain generalization check.
400 paired prompts JSONL text Thai English
Apr 28, 2026 View on HF