List your company
ResearchFormerly WizWand

Natural Language Graph Reasoning on NLGraph (val)

54.5AccuracyEvoGM
26.4233.714148.29May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
EvoGMBackbone=Qwen2.5-1.5B2026.0554.5
PSO-MergingBackbone=Qwen2.5-1.5B2026.0545.5
CMABackbone=Qwen2.5-1.5B2026.0543
TIESBase model=Qwen2.5-1.5B2026.0539
TIESBackbone=Qwen2.5-1.5B2026.0539
PSO-MergingBase model=Qwen2.5-1.5B2026.0538
Model SwarmBackbone=Qwen2.5-1.5B2026.0536
Single BestBase model=Qwen2.5-1.5B2026.0535.5
Single BestBackbone=Qwen2.5-1.5B2026.0535.5
MTLBase model=Qwen2.5-1.5B2026.0532
MTLBackbone=Qwen2.5-1.5B2026.0532
Model SwarmBase model=Qwen2.5-1.5B2026.0530.5
EvoGMBase model=Qwen2.5-1.5B2026.0529.5
BaseBase model=Qwen2.5-1.5B2026.0528.5
Model SoupBase model=Qwen2.5-1.5B2026.0528.5
DAREBase model=Qwen2.5-1.5B2026.0528.5
BaseBackbone=Qwen2.5-1.5B2026.0528.5
Model SoupBackbone=Qwen2.5-1.5B2026.0528.5
DAREBackbone=Qwen2.5-1.5B2026.0528.5
Task ArithmeticBase model=Qwen2.5-1.5B2026.0528
TABackbone=Qwen2.5-1.5B2026.0528
CMABase model=Qwen2.5-1.5B2026.0527.5