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 TitleRepository
Golden Transformer0.917--
Human Benchmark0.915RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark
ruRoberta-large finetune0.82--
ruBert-large finetune0.773--
ruT5-base-finetune0.732--
ruBert-base finetune0.712--
ruT5-large-finetune0.711--
SBERT_Large_mt_ru_finetuning0.697--
SBERT_Large0.675--
MT5 Large0.657mT5: A massively multilingual pre-trained text-to-text transformer
heuristic majority0.642Unreasonable Effectiveness of Rule-Based Heuristics in Solving Russian SuperGLUE Tasks-
RuBERT plain0.639--
YaLM 1.0B few-shot0.637--
RuGPT3Medium0.634--
Multilingual Bert0.624--
Baseline TF-IDF1.10.621RussianSuperGLUE: A Russian Language Understanding Evaluation Benchmark
RuGPT3Small0.61--
RuBERT conversational0.606--
RuGPT3Large0.604--
RuGPT3XL few-shot0.59--
0 of 22 row(s) selected.
Question Answering On Danetqa | SOTA | HyperAI超神经