HyperAI超神経

Open Domain Dialog On Kilt Wizard Of

評価指標

F1
KILT-F1
KILT-RL
R-Prec
ROUGE-L
Recall@5

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

モデル名
F1
KILT-F1
KILT-RL
R-Prec
ROUGE-L
Recall@5
Paper TitleRepository
chriskuei0.00.00.064.790.082.15--
bart-base14.820.00.00.013.350.0--
aa_evalai17.30.00.00.015.930.0--
GENRE0.00.00.062.880.077.74--
multitask3.092.182.0455.712.9275.59--
KGI18.5711.7910.3655.3716.3678.45--
Sphere17.280.00.00.015.710.0--
intersect18.3411.6310.4557.5516.6578.96--
TABi0.00.00.059.110.069.1--
Hindsight19.1913.3911.9256.0817.0674.27--
Wikipedia15.667.576.5541.5413.9468.25--
BART12.860.00.00.011.770.0--
T5-base13.530.00.00.012.40.0KILT: a Benchmark for Knowledge Intensive Language Tasks
Multitask DPR + BART15.126.965.9141.0613.2767.13--
Routing Transformer, c-REALM12.154.84.4139.0611.4251.63--
Re2G18.912.9811.3960.116.7679.98Re2G: Retrieve, Rerank, Generate
multi-task small13.750.00.00.012.810.0--
TransMemNet11.852.21.8518.3510.1118.35--
RAG13.118.757.5957.7511.5774.61--
BART + DPR15.194.373.7125.4613.2351.19--
0 of 21 row(s) selected.