List your company

Single-Hop Question Answering on LoCoMo Single-Hop (test)

37.9F1Reproduced Baselines
7.708815.546923.38531.2231Feb 17, 2026

Evaluation results

MethodLinks
Reproduced BaselinesModel=GPT-oss:20b, Memory=A-mem2026.0237.932.880
PoisonedRAGModel=GPT-oss:20b, Memory=A-mem2026.0235.1430.36-7.2
Reproduced BaselinesModel=Gemma3:27b, Memory=Mem02026.0230.8427.420
Reproduced BaselinesModel=Gemma3:27b, Memory=A-mem2026.0230.826.280
Reproduced BaselinesModel=Llama3.2:3b, Memory=A-mem2026.0230.3250
PoisonedRAGModel=Gemma3:27b, Memory=Mem02026.0229.0325.74-6.1
PoisonedRAGModel=Gemma3:27b, Memory=A-mem2026.0228.8324.43-13.4
PoisonedRAGModel=Llama3.2:3b, Memory=A-mem2026.0226.2721.64-7.1
PoisonedRAGModel=GPT-oss:20b, Memory=Mem02026.0226.1822.7-13.1
Reproduced BaselinesModel=GPT-oss:20b, Memory=Mem02026.0225.9322.560
Reproduced BaselinesModel=Llama3.2:3b, Memory=Mem02026.0225.4622.010
PoisonedRAGModel=Llama3.2:3b, Memory=Mem02026.0224.4921.15-4.2
ER-MIAModel=GPT-oss:20b, Attack Size=1 Memory / Question, Memory=A-mem2026.0219.3216.88-48.4
ER-MIAModel=GPT-oss:20b, Attack Size=2 Memories / Question, Memory=A-mem2026.0216.6614.25-55.5
ER-MIAModel=Gemma3:27b, Attack Size=1 Memory / Question, Memory=A-mem2026.021512.48-46.5
ER-MIAModel=Llama3.2:3b, Attack Size=1 Memory / Question, Memory=A-mem2026.0214.9912.28-40.2
ER-MIAModel=Gemma3:27b, Attack Size=1 Memory / Question, Memory=Mem02026.0213.4611.96-48.4
ER-MIAModel=Llama3.2:3b, Attack Size=1 Memory / Question, Memory=Mem02026.0212.510.7-44.1
ER-MIAModel=Gemma3:27b, Attack Size=2 Memories / Question, Memory=Mem02026.0211.810.51-53.6
ER-MIAModel=GPT-oss:20b, Attack Size=1 Memory / Question, Memory=Mem02026.0211.119.63-51.6
ER-MIAModel=Llama3.2:3b, Attack Size=2 Memories / Question, Memory=A-mem2026.0210.648.38-46.8
ER-MIAModel=Gemma3:27b, Attack Size=2 Memories / Question, Memory=A-mem2026.0210.48.53-61.6
ER-MIAModel=GPT-oss:20b, Attack Size=2 Memories / Question, Memory=Mem02026.029.688.32-56.7
ER-MIAModel=Llama3.2:3b, Attack Size=2 Memories / Question, Memory=Mem02026.028.877.81-53