Continual Learning on Multi-hop QA Domain-level stream
52.31AA ScoreMasFACT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MasFACTReplay-based=false2026.05 | 52.31 | -0.55 | |
| MasFACTReplay-based=false2026.05 | 52.31 | -0.55 | |
| ARGDesignerReplay-based=true2026.05 | 49.32 | 1.9 | |
| ARGDesignerReplay-based=true2026.05 | 49.32 | 1.9 | |
| MasRouterReplay-based=true2026.05 | 48.79 | 0.72 | |
| MasRouterReplay-based=true2026.05 | 48.79 | 0.72 | |
| GDesignerReplay-based=true2026.05 | 47.97 | 1.26 | |
| GDesignerReplay-based=true2026.05 | 47.97 | 1.26 | |
| AgentDropoutReplay-based=true2026.05 | 47.17 | 1.04 | |
| AgentDropoutReplay-based=true2026.05 | 47.17 | 1.04 | |
| ARGDesignerReplay-based=false2026.05 | 46.32 | 4.27 | |
| ARGDesignerReplay-based=false2026.05 | 46.32 | 4.27 | |
| MasRouterReplay-based=false2026.05 | 45.26 | 3.74 | |
| MasRouterReplay-based=false2026.05 | 45.26 | 3.74 | |
| GDesignerReplay-based=false2026.05 | 44.92 | 5.23 | |
| GDesignerReplay-based=false2026.05 | 44.92 | 5.23 | |
| AFLOWReplay-based=true2026.05 | 43.68 | 1.89 | |
| AFLOWReplay-based=true2026.05 | 43.68 | 1.89 | |
| AgentDropoutReplay-based=false2026.05 | 42.94 | 5.19 | |
| AgentDropoutReplay-based=false2026.05 | 42.94 | 5.19 | |
| SC(CoT)-ICLReplay-based=false2026.05 | 41.51 | 3.24 | |
| SC(CoT)-ICLReplay-based=false2026.05 | 41.51 | 3.24 | |
| AFLOWReplay-based=false2026.05 | 40.18 | 4.72 | |
| AFLOWReplay-based=false2026.05 | 40.18 | 4.72 | |
| CompleteReplay-based=false2026.05 | 39.62 | — | |
| Complete GraphReplay-based=false2026.05 | 39.62 | — | |
| CoT-ICLReplay-based=false2026.05 | 38.74 | 4.76 | |
| CoT-ICLReplay-based=false2026.05 | 38.74 | 4.76 | |
| TreeReplay-based=false2026.05 | 37.81 | — | |
| TreeReplay-based=false2026.05 | 37.81 | — | |
| VanillaReplay-based=false2026.05 | 35.85 | — | |
| VanillaReplay-based=false2026.05 | 35.85 | — | |
| ChainReplay-based=false2026.05 | 32.03 | — | |
| ChainReplay-based=false2026.05 | 32.03 | — |