List your company

Model Editing on WikiBigEdit (Efficacy, Specificity, Personas, Reasoning)

79.6EfficacyUltraEdit
-3.18418.30839.861.292May 20, 2025

Evaluation results

MethodLinks
UltraEditBackbone=LLaMA-3-8B-Instruct, Edit Scenario=standard2025.0579.673.4948.5166.5535.64
UltraEditBackbone=LLaMA-3-8B-Instruct2025.0579.673.4948.5166.5535.64
UltraEditBackbone=Mistral-7B-v0.3, Edit Scenario=standard2025.057670.1546.0962.2735.8
UltraEditBackbone=Mistral-7B-v0.32025.057670.1546.0962.2735.8
RLEditBackbone=LLaMA-3-8B-Instruct, Edit Scenario=standard2025.0575.357037.2165.5528.13
UltraEditBackbone=GPT-J, Edit Scenario=standard2025.0573.8466.5737.1756.929.27
UltraEditBackbone=GPT-J2025.0573.8466.5737.1756.929.27
UltraEditModel=Qwen2.5-7B-Instruct2025.0573.3765.8645.1254.6532.74
UltraEdit*Backbone=Mistral-7B-v0.3, Edit Scenario=ultra-large-scale2025.0571.7865.6355.456.11—
UltraEdit*Backbone=Mistral-7B-v0.32025.0571.7865.6355.456.11—
AlphaEditBackbone=GPT-J, Edit Scenario=standard2025.0569.6655.0321.242.60.07
UltraEdit*Backbone=LLaMA-3-8B-Instruct, Edit Scenario=ultra-large-scale2025.0568.9963.5952.2855.04—
UltraEdit*Backbone=LLaMA-3-8B-Instruct2025.0568.9963.5952.2855.04—
UltraEdit*Backbone=GPT-J, Edit Scenario=ultra-large-scale2025.0566.4660.5447.951.73—
UltraEdit*Backbone=GPT-J2025.0566.4660.5447.951.73—
UltraEdit*Model=Qwen2.5-7B-Instruct2025.0566.3460.4251.7450.53—
RLEditBackbone=GPT-J, Edit Scenario=standard2025.0566.1860.9732.255.7825.94
AlphaEditBackbone=LLaMA-3-8B-Instruct, Edit Scenario=standard2025.0563.2454.6820.1742.580.01
RLEditBackbone=Mistral-7B-v0.3, Edit Scenario=standard2025.0557.5552.4728.7849.2122.41
WISEBackbone=GPT-J, Edit Scenario=standard2025.0549.8845.393040.5221.01
WISEBackbone=Mistral-7B-v0.3, Edit Scenario=standard2025.0537.3133.4411.6127.448.95
WISEBackbone=LLaMA-3-8B-Instruct, Edit Scenario=standard2025.0534.0732.2628.9129.5521.19
FTBackbone=GPT-J, Edit Scenario=standard2025.0521.917.568.4713.919.24
AlphaEditModel=Qwen2.5-7B-Instruct2025.0520.3115.492.179.010.23
FTBackbone=Mistral-7B-v0.3, Edit Scenario=standard2025.0513.7714.8611.8411.557.51
FTBackbone=LLaMA-3-8B-Instruct, Edit Scenario=standard2025.051311.76.7511.024.38
FTModel=Qwen2.5-7B-Instruct2025.0510.357.593.685.555.84
RLEditModel=Qwen2.5-7B-Instruct2025.052.831.810.431.450.39
PRUNEBackbone=GPT-J2025.052.362.321.011.860
RECTModel=Qwen2.5-7B-Instruct2025.052.342.290.830.780
PRUNEModel=Qwen2.5-7B-Instruct2025.052.342.340.971.370
RECTBackbone=GPT-J2025.052.131.990.592.060
MEMITBackbone=GPT-J2025.051.591.590.50.80
MEMITModel=Qwen2.5-7B-Instruct2025.050.540.330.380.320
RECTBackbone=LLaMA-3-8B-Instruct2025.050.210.240.060.290
MALMENModel=Qwen2.5-7B-Instruct2025.050.060.020.0200
MENDBackbone=GPT-J2025.050.020.010.020.020.02
MEMITBackbone=LLaMA-3-8B-Instruct2025.050.020.020.090.020
PRUNEBackbone=LLaMA-3-8B-Instruct2025.050.020.020.090.020
MENDBackbone=Mistral-7B-v0.32025.050.01000.010
MENDBackbone=LLaMA-3-8B-Instruct2025.050.010.010.10.010
AlphaEditBackbone=Mistral-7B-v0.3, Edit Scenario=standard2025.0500000
MENDModel=Qwen2.5-7B-Instruct2025.0500000
MALMENBackbone=GPT-J2025.0500.010.0100
MEMITBackbone=Mistral-7B-v0.32025.0500000
MALMENBackbone=Mistral-7B-v0.32025.05000.0100
RECTBackbone=Mistral-7B-v0.32025.0500000
PRUNEBackbone=Mistral-7B-v0.32025.0500000
MALMENBackbone=LLaMA-3-8B-Instruct2025.0500000