List your company

Model Editing on ZsRE (Reliability, Generalization, Locality, Overall Performance)

1ReliabilityHoReN
-0.040.230.50.77May 2, 2026

Evaluation results

MethodLinks
HoReNModel=LLaMA-3 8B, Editing Scale (N)=1002026.0510.8110.93
HoReNModel=LLaMA-3 8B, Editing Scale (N)=5002026.0510.8410.94
GRACEModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.0510.0610.39
GRACEModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.0510.0510.37
HoReNModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.0510.9210.97
GRACEModel=Qwen-2.5 7B, Editing Scale (N)=1002026.0510.0210.27
GRACEModel=Qwen-2.5 7B, Editing Scale (N)=5002026.0510.0210.27
HoReNModel=Qwen-2.5 7B, Editing Scale (N)=5002026.0510.930.990.97
GRACEModel=Qwen-2.5 7B, Editing Scale (N)=10002026.0510.0210.27
HoReNModel=Qwen-2.5 7B, Editing Scale (N)=10002026.0510.930.980.97
HoReNModel=SEA-LION 32B, Editing Scale (N)=1002026.0510.9110.97
HoReNModel=SEA-LION 32B, Editing Scale (N)=5002026.0510.90.990.96
HoReNModel=SEA-LION 32B, Editing Scale (N)=10002026.0510.90.980.96
GRACEModel=LLaMA-3 8B, Editing Scale (N)=1002026.050.990.0110.21
GRACEModel=LLaMA-3 8B, Editing Scale (N)=5002026.050.990.0210.27
AlphaEditModel=LLaMA-3 8B, Editing Scale (N)=5002026.050.990.940.830.92
GRACEModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.990.0210.27
HoReNModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.990.8410.94
HoReNModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.050.990.8810.96
GRACEModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.990.0510.37
HoReNModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.990.930.990.97
HoReNModel=Qwen-2.5 7B, Editing Scale (N)=1002026.050.990.9310.97
AlphaEditModel=Qwen-2.5 7B, Editing Scale (N)=10002026.050.990.880.880.92
AlphaEditModel=LLaMA-3 8B, Editing Scale (N)=1002026.050.980.920.890.93
AlphaEditModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.980.920.780.89
AlphaEditModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.050.980.90.880.92
AlphaEditModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.050.980.930.80.9
AlphaEditModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.970.920.740.87
AlphaEditModel=SEA-LION 32B, Editing Scale (N)=5002026.050.970.770.960.89
AlphaEditModel=SEA-LION 32B, Editing Scale (N)=10002026.050.970.770.950.89
AlphaEditModel=SEA-LION 32B, Editing Scale (N)=1002026.050.960.760.990.9
WISEModel=SEA-LION 32B, Editing Scale (N)=1002026.050.940.7210.88
UltraEditModel=Qwen-2.5 7B, Editing Scale (N)=1002026.050.920.80.990.9
UltraEditModel=LLaMA-3 8B, Editing Scale (N)=1002026.050.910.880.690.82
UltraEditModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.050.910.890.690.82
UltraEditModel=LLaMA-3 8B, Editing Scale (N)=5002026.050.90.880.560.76
WISEModel=Qwen-2.5 7B, Editing Scale (N)=1002026.050.90.8710.92
UltraEditModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.050.890.870.610.78
UltraEditModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.880.860.540.74
UltraEditModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.870.850.540.74
WISEModel=Qwen-2.5 7B, Editing Scale (N)=5002026.050.870.80.80.82
UltraEditModel=Qwen-2.5 7B, Editing Scale (N)=5002026.050.870.810.960.88
AlphaEditModel=Qwen-2.5 7B, Editing Scale (N)=1002026.050.860.80.980.88
AlphaEditModel=Qwen-2.5 7B, Editing Scale (N)=5002026.050.850.790.910.85
WISEModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.050.830.7910.87
UltraEditModel=Qwen-2.5 7B, Editing Scale (N)=10002026.050.80.750.890.81
WISEModel=SEA-LION 32B, Editing Scale (N)=5002026.050.80.6810.82
UltraEditModel=SEA-LION 32B, Editing Scale (N)=1002026.050.760.70.980.8
UltraEditModel=SEA-LION 32B, Editing Scale (N)=5002026.050.750.720.950.8
UltraEditModel=SEA-LION 32B, Editing Scale (N)=10002026.050.720.690.930.77
WISEModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.050.710.6610.78
WISEModel=SEA-LION 32B, Editing Scale (N)=10002026.050.70.6310.76
WISEModel=Qwen-2.5 7B, Editing Scale (N)=10002026.050.630.580.60.6
WISEModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.620.590.990.71
WISEModel=LLaMA-3 8B, Editing Scale (N)=5002026.050.560.5410.67
WISEModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.550.540.990.66
WISEModel=LLaMA-3 8B, Editing Scale (N)=1002026.050.540.510.65
ROMEModel=Qwen-2.5 7B, Editing Scale (N)=1002026.050.510.510.320.44
GRACEModel=SEA-LION 32B, Editing Scale (N)=1002026.050.460.0210.21
GRACEModel=SEA-LION 32B, Editing Scale (N)=10002026.050.430.0210.21
GRACEModel=SEA-LION 32B, Editing Scale (N)=5002026.050.410.0210.2
ROMEModel=Qwen-2.5 7B, Editing Scale (N)=5002026.050.30.280.060.17
ROMEModel=Qwen-2.5 7B, Editing Scale (N)=10002026.050.190.190.020.09
ROMEModel=LLaMA-3 8B, Editing Scale (N)=1002026.050.170.160.010.06
ROMEModel=LLaMA-3.1 8B, Editing Scale (N)=1002026.050.040.050.010.03
ROMEModel=LLaMA-3.1 8B, Editing Scale (N)=10002026.050.030.020.030.03
ROMEModel=LLaMA-3 8B, Editing Scale (N)=5002026.050.020.0300
ROMEModel=LLaMA-3 8B, Editing Scale (N)=10002026.050.020.0200
ROMEModel=LLaMA-3.1 8B, Editing Scale (N)=5002026.050.020.020.020.02
ROMEModel=SEA-LION 32B, Editing Scale (N)=1002026.050.020.020.020.02
ROMEModel=SEA-LION 32B, Editing Scale (N)=5002026.050000
ROMEModel=SEA-LION 32B, Editing Scale (N)=10002026.050000