SplaXBERT通过混合精度与分段上下文提升长文本问答效率

SplaXBERT: Leveraging Mixed Precision Training and Context Splitting for Question Answering

• 采用上下文分段与混合精度训练策略
• 在SQuAD v1.1上达85.95%精确匹配率和92.97%F1值
• 兼顾准确率与资源效率，适合长文本场景

SplaXBERT基于ALBERT-xlarge，结合上下文分段与混合精度训练，在长文本问答任务中实现高效率。在SQuAD v1.1数据集上，其精确匹配率为85.95%，F1得分为92.97%，在准确率和资源效率方面均优于传统BERT类模型。

原文摘要 · Abstract (English)

SplaXBERT, built on ALBERT-xlarge with context-splitting and mixed precision training, achieves high efficiency in question-answering tasks on lengthy texts. Tested on SQuAD v1.1, it attains an Exact Match of 85.95% and an F1 Score of 92.97%, outperforming traditional BERT-based models in both accuracy and resource efficiency.

Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。