Question Answering On Danetqa
评估指标
Accuracy
评测结果
各个模型在此基准测试上的表现结果
| Paper Title | Repository | ||
|---|---|---|---|
| Golden Transformer | 0.917 | - | - |
| Human Benchmark | 0.915 | RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark | |
| ruRoberta-large finetune | 0.82 | - | - |
| ruBert-large finetune | 0.773 | - | - |
| ruT5-base-finetune | 0.732 | - | - |
| ruBert-base finetune | 0.712 | - | - |
| ruT5-large-finetune | 0.711 | - | - |
| SBERT_Large_mt_ru_finetuning | 0.697 | - | - |
| SBERT_Large | 0.675 | - | - |
| MT5 Large | 0.657 | mT5: A massively multilingual pre-trained text-to-text transformer | |
| heuristic majority | 0.642 | Unreasonable Effectiveness of Rule-Based Heuristics in Solving Russian SuperGLUE Tasks | - |
| RuBERT plain | 0.639 | - | - |
| YaLM 1.0B few-shot | 0.637 | - | - |
| RuGPT3Medium | 0.634 | - | - |
| Multilingual Bert | 0.624 | - | - |
| Baseline TF-IDF1.1 | 0.621 | RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark | |
| RuGPT3Small | 0.61 | - | - |
| RuBERT conversational | 0.606 | - | - |
| RuGPT3Large | 0.604 | - | - |
| RuGPT3XL few-shot | 0.59 | - | - |
0 of 22 row(s) selected.