في الإجابة على الأسئلة المفتوحة للنطاق، أصبح استرجاع المقطع الكثيف نموذجا جديدا لاسترداد الممرات ذات الصلة لإيجاد الإجابات. عادة ما يتم اعتماد بنية التشفير المزدوجة لتعلم تمثيلات كثيفة من الأسئلة والممرات للمطابقة الدلالية. ومع ذلك، من الصعب تدريب تشفير مزدوج بشكل فعال بسبب التحديات بما في ذلك التناقض بين التدريب والاستدلال، ووجود إيجابيات غير محدودة وبيانات تدريب محدودة. لمعالجة هذه التحديات، نقترح نهج تدريبي محسن، يسمى Rocketqa، لتحسين استرجاع الممر الكثيف. نجعل ثلاث مساهمات تقنية رئيسية في Rocketqa، وهي السلبيات عبر الدفعة، السلبيات الصلبة الشاقة وزعم البيانات. تظهر نتائج التجربة أن Rocketqa تتفوق بشكل كبير على النماذج السابقة من بين الفنادق السابقة على كل من MSMARCO والأسئلة الطبيعية. نقوم أيضا بإجراء تجارب مكثفة لفحص فعالية الاستراتيجيات الثلاث في Rocketqa. علاوة على ذلك، نوضح أن أداء ضمان الجودة المناسبة يمكن تحسينه بناء على مسترد Rocketqa لدينا.
In open-domain question answering, dense passage retrieval has become a new paradigm to retrieve relevant passages for finding answers. Typically, the dual-encoder architecture is adopted to learn dense representations of questions and passages for semantic matching. However, it is difficult to effectively train a dual-encoder due to the challenges including the discrepancy between training and inference, the existence of unlabeled positives and limited training data. To address these challenges, we propose an optimized training approach, called RocketQA, to improving dense passage retrieval. We make three major technical contributions in RocketQA, namely cross-batch negatives, denoised hard negatives and data augmentation. The experiment results show that RocketQA significantly outperforms previous state-of-the-art models on both MSMARCO and Natural Questions. We also conduct extensive experiments to examine the effectiveness of the three strategies in RocketQA. Besides, we demonstrate that the performance of end-to-end QA can be improved based on our RocketQA retriever.
References used
https://aclanthology.org/
Dense neural text retrieval has achieved promising results on open-domain Question Answering (QA), where latent representations of questions and passages are exploited for maximum inner product search in the retrieval process. However, current dense
In various natural language processing tasks, passage retrieval and passage re-ranking are two key procedures in finding and ranking relevant information. Since both the two procedures contribute to the final performance, it is important to jointly o
Dense retrieval has shown great success for passage ranking in English. However, its effectiveness for non-English languages remains unexplored due to limitation in training resources. In this work, we explore different transfer techniques for docume
Recent advances in open-domain QA have led to strong models based on dense retrieval, but only focused on retrieving textual passages. In this work, we tackle open-domain QA over tables for the first time, and show that retrieval can be improved by a
In the context of neural passage retrieval, we study three promising techniques: synthetic data generation, negative sampling, and fusion. We systematically investigate how these techniques contribute to the performance of the retrieval system and ho