Bidirectional Attention Flow for Machine Comprehension

5 Nov 2016Minjoon SeoAniruddha KembhaviAli FarhadiHannaneh Hajishirzi

Machine comprehension (MC), answering a query about a given context paragraph, requires modeling complex interactions between the context and the query. Recently, attention mechanisms have been successfully extended to MC... (read more)

PDF Abstract

Evaluation results from the paper


Task Dataset Model Metric name Metric value Global rank Compare
Question Answering CNN / Daily Mail BiDAF CNN 76.9 # 4
Question Answering CNN / Daily Mail BiDAF Daily Mail 79.6 # 2
Question Answering MS MARCO BiDaF Baseline Rouge-L 23.96 # 4
Question Answering MS MARCO BiDaF Baseline BLEU-1 10.64 # 4
Question Answering NarrativeQA BiDAF BLEU-1 33.45 # 7
Question Answering NarrativeQA BiDAF BLEU-4 15.69 # 7
Question Answering NarrativeQA BiDAF METEOR 15.68 # 7
Question Answering NarrativeQA BiDAF Rouge-L 36.74 # 7
Open-Domain Question Answering Quasar BiDAF EM (Quasar-T) 25.9 # 6
Open-Domain Question Answering Quasar BiDAF F1 (Quasar-T) 28.5 # 5
Question Answering SQuAD1.1 BiDAF (single model) EM 67.974 # 126
Question Answering SQuAD1.1 BiDAF (single model) F1 77.323 # 128
Question Answering SQuAD1.1 BiDAF (ensemble) EM 73.744 # 91
Question Answering SQuAD1.1 BiDAF (ensemble) F1 81.525 # 95