ChineseGLUE — leaderboard
Original ChineseGLUE language-understanding benchmark for Chinese classification, sentence-pair, NER, and reading-comprehension tasks.
Metric: Classification Score (%). Source: github.com. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | RoBERTa-wwm-large | 85.13 |
| 2 | RoBERTa-large | 85.08 |
| 3 | BERT-wwm-ext | 84.89 |
| 4 | ERNIE-base | 84.63 |
| 5 | BERT-base | 84.57 |
| 6 | RoBERTa-wwm-ext | 84.55 |
| 7 | ALBERT-xlarge | 84.08 |
| 8 | XLNet-mid | 81.07 |
| 9 | ALBERT-tiny | 78.22 |
Interactive version: theaggregate.ai/benchmark?slug=chineseglue · How the rankings work · Data refreshed daily, snapshot 2026-07-22.