sbintuitions/JMTEB
JMTEB: Japanese Massive Text Embedding Benchmark JMTEB is a benchmark for evaluating Japanese text embedding models. It consists of 5 tasks, currently involving 28 datasets in total. You can find the update history here. TL;DR from datasets import load_dataset dataset = load_dataset("sbintuitions/JMTEB", name="<dataset_name>", split="<split>") JMTEB_DATASET_NAMES = ( 'livedoor_news', 'mewsc16_ja', 'sib200_japanese_clustering'… See the full description on the dataset page: https://huggingface.co/datasets/sbintuitions/JMTEB.
| Author / organization | sbintuitions |
|---|---|
| License (as declared) | cc-by-sa-4.0 |
| Downloads | 2.6K (as reported by Hugging Face) |
| Likes | 19 |
| Created | 11 March 2024 |
| Last updated at source | 16 March 2026 |
Popularity, last 3 days
2,646 now (+469 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
More from sbintuitions
- sbintuitions/JamC-QADataset
- sbintuitions/JCommonsenseQADataset
- sbintuitions/WildGuardTestJPDataset
- sbintuitions/JSQuADDataset
- sbintuitions/JMTEB-liteDataset
- sbintuitions/joyo-kanji-yomi-benchmarkDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.