Uni-Encoder: A Fast and Accurate Response Selection Paradigm for Generation-Based Dialogue Systems

Benchmark Model Rank Results
conversational-response-selection-on-douban-1Uni-Enc+BERT-FP#2MAP: 0.648MRR: 0.688P@1: 0.518R10@1: 0.327R10@2: 0.557
conversational-response-selection-on-douban-1Uni-Encoder#7MAP: 0.622MRR: 0.662P@1: 0.481R10@1: 0.303R10@2: 0.514
conversational-response-selection-on-ubuntu-1Uni-Enc+BERT-FP#3R10@1: 0.916R10@2: 0.965R10@5: 0.994
conversational-response-selection-on-ubuntu-1Uni-Encoder#5R10@1: 0.886R10@2: 0.946R10@5: 0.989