Dataset · shunk031
JGLUE Japanese language understanding benchmark
shunk031/JGLUE
Japanese-language tasks for evaluating model text comprehension capabilities.
What's inside
The benchmark was created from scratch in Japanese without using machine or manual translation from other languages. It includes tasks for text classification, question answering, multiple choice, and sentence semantic similarity evaluation.
What you can do with it
- Evaluating Japanese language comprehension quality in LLMs and encoders
- Fine-tuning models for text classification and semantic similarity search in Japanese
- Testing question answering (QA) systems for the Japanese segment
- Tasks
- Multiple choice, Question answering, Sentence similarity, Text classification
- Languages
- Japanese
- Downloads this month
- 2.5K
- Likes
- 48
Models trained on it
- hotchpotch/japanese-reranker-cross-encoder-small-v1Text ranking 111.5K
- pkshatech/GLuCoSE-base-jaSentence similarity 31.8K
- hotchpotch/japanese-reranker-cross-encoder-xsmall-v1Text ranking 20.3K
- hotchpotch/japanese-reranker-cross-encoder-base-v1Text ranking 1.2K
- hotchpotch/japanese-reranker-cross-encoder-large-v1Text ranking 1.1K
- akiFQC/bert-base-japanese-v3_nli-jsnli-jnli-jsickZero shot classification 1.1K
- hotchpotch/japanese-bge-reranker-v2-m3-v1Text ranking 976
- bclavie/fio-base-japanese-v0.1Sentence similarity 149
How to load it
from datasets import load_dataset
ds = load_dataset("shunk031/JGLUE")