Factoid Knowledge Recall on PopQA Factoid
86.6ID ScoreREMIX (- / K-Pile)
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| REMIX (- / K-Pile)Backbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=- / K-Pile2024.11 | 86.6 | 90.8 | 93.9 | 74.4 | 86.4 | |
| REMIX (Random / K-Pile)Backbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=Random / K-Pile2024.11 | 82.6 | 85.8 | 90.7 | 80.5 | 84.9 | |
| REMIX (Random / -)Backbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=Random / -2024.11 | 62 | 17.7 | 69.3 | 65.8 | 53.7 | |
| No MixingBackbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=No Mixing2024.11 | 49.8 | 7.7 | 57.8 | 72.5 | 47 | |
| REMIX (- / Random)Backbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=- / Random2024.11 | 35.7 | 5.2 | 38.1 | 45.9 | 31.2 | |
| REMIX (K-Pile / -)Backbone=Llama-3-8B, Mixing Strategy (DM1 / DM2)=K-Pile / -2024.11 | 24 | 2.8 | 11.3 | 31.8 | 17.5 |