List your company

Model Editing on FEVER 20K edits (test)

99.07EfficacySTABLEEDIT
-3.962822.786149.53576.2839May 12, 2026

Evaluation results

MethodLinks
STABLEEDITBackbone=Qwen2.5-7B-Instruct2026.0599.0794.8769.61
STABLEEDITBackbone=Mistral-7B-v0.32026.0598.3896.5583.89
STABLEEDITBackbone=GPT-J-6B2026.0598.3596.9979.85
ULTRAEDITBackbone=Qwen2.5-7B-Instruct2026.0597.9793.9168.86
STABLEEDITBackbone=Llama-3-8B-Instruct2026.0597.8994.4769.24
ULTRAEDITBackbone=Mistral-7B-v0.32026.0597.8796.0984.29
ULTRAEDITBackbone=GPT-J-6B2026.0597.4596.3779.72
RLEDITBackbone=GPT-J-6B2026.0597.2996.0179.86
RLEDITBackbone=Qwen2.5-7B-Instruct2026.0596.9393.9268.28
ULTRAEDITBackbone=Llama-3-8B-Instruct2026.0595.3991.9367.14
MALMENBackbone=Llama-3-8B-Instruct2026.0595.0392.4469.2
RLEDITBackbone=Mistral-7B-v0.32026.0592.3591.3971.85
RLEDITBackbone=Llama-3-8B-Instruct2026.0591.3889.9341.98
MENDBackbone=Qwen2.5-7B-Instruct2026.0576.4377.6640.39
MENDBackbone=GPT-J-6B2026.0552.851.4445.42
MENDBackbone=Llama-3-8B-Instruct2026.0536.1936.1924.31
ALPHAEDITBackbone=Qwen2.5-7B-Instruct2026.0532.7831.1922.12
FTBackbone=Qwen2.5-7B-Instruct2026.0526.0924.6221.36
FTBackbone=Mistral-7B-v0.32026.0523.823.3716.3
MALMENBackbone=Mistral-7B-v0.32026.0518.4217.4312.09
FTBackbone=Llama-3-8B-Instruct2026.0516.2113.085.01
FTBackbone=GPT-J-6B2026.0514.0213.989.26
Pre-editedBackbone=GPT-J-6B2026.059.619.6815.9
ALPHAEDITBackbone=Llama-3-8B-Instruct2026.056.396.142.72
MEMITBackbone=GPT-J-6B2026.055.545.035.46
ALPHAEDITBackbone=GPT-J-6B2026.051.891.871.85
MALMENBackbone=GPT-J-6B2026.051.331.252.92
Pre-editedBackbone=Qwen2.5-7B-Instruct2026.050.570.62.17
Pre-editedBackbone=Mistral-7B-v0.32026.050.410.51.98
MEMITBackbone=Qwen2.5-7B-Instruct2026.050.080.120.15
MALMENBackbone=Qwen2.5-7B-Instruct2026.050.060.060.07
Pre-editedBackbone=Llama-3-8B-Instruct2026.050.020.020.26
MEMITBackbone=Llama-3-8B-Instruct2026.050.020.020.02
MEMITBackbone=Mistral-7B-v0.32026.05000
ALPHAEDITBackbone=Mistral-7B-v0.32026.05000
MENDBackbone=Mistral-7B-v0.32026.05000