List your company

Vision-Language Model Editing on FVQA 1.0 (test)

100AccuracyReasonEdit
-4235077Feb 2, 2026

Evaluation results

MethodLinks
ReasonEditVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.021000.970.950.970.87100
GRACEVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02980.690.920.540.7699
GRACE-COTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02980.690.920.540.7699
GRACEVLM Backbone=InstructBLIP-7B, Error Count=7642026.02980.590.910.560.899
GRACE-COTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02980.570.910.570.898
GRACEVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02980.790.790.60.7586
GRACE-COTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02980.790.790.60.7586
GRACEVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02980.770.820.60.7393
GRACE-COTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02980.770.820.60.7393
ReasonEditVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02980.850.910.950.83100
ReasonEditVLM Backbone=InstructBLIP-7B, Error Count=7642026.02960.940.920.920.88100
ReasonEditVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02960.850.920.950.81100
IKEVLM Backbone=InstructBLIP-7B, Error Count=7642026.02930.870.950.860.75100
IKE-COTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02930.850.940.840.8100
IKEVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02920.80.970.70.78100
IKE-COTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02920.780.90.770.79100
IKE-COTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02910.730.870.690.78100
IKEVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02900.760.860.70.75100
IKE-COTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02880.790.860.850.81100
IKEVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02870.80.860.870.79100
BalancEdit-COTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02860.80.60.570.671
FT-COTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02850.770.750.710.6166
BalancEditVLM Backbone=InstructBLIP-7B, Error Count=7642026.02850.620.640.470.5560
FT-COTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02840.680.790.610.6771
FTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02830.760.690.70.3953
BalancEditVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02830.790.530.470.4749
FTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02780.640.680.560.457
FT-COTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02780.710.620.660.548
BalancEditVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02730.710.380.450.3239
FTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02690.640.530.580.2738
FT-COTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02680.650.570.620.4949
BalancEditVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02660.640.40.410.3231
FTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02640.60.510.580.2736
BalancEdit-COTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02550.490.610.610.7388
MENDVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02520.50.450.490.2640
MEND-COTVLM Backbone=InstructBLIP-7B, Error Count=7642026.02500.440.580.450.5166
BalancEdit-COTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02460.450.430.410.5752
MEND-COTVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.02450.410.530.450.4959
MENDVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02440.420.370.410.229
BalancEdit-COTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02400.40.390.410.642
MEND-COTVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02380.380.370.340.3438
MEND-COTVLM Backbone=Qwen3-VL-4B, Error Count=7002026.02360.330.290.330.3332
MENDVLM Backbone=Qwen3-VL-8B, Error Count=8572026.02330.330.290.290.1927
MENDVLM Backbone=InstructBLIP-7B, Error Count=7642026.02280.250.270.260.2234
UneditedVLM Backbone=LLaVA-1.5-7B, Error Count=8112026.0200.170.520.450.76100
UneditedVLM Backbone=InstructBLIP-7B, Error Count=7642026.0200.250.520.540.8100
UneditedVLM Backbone=Qwen3-VL-8B, Error Count=8572026.0200.160.470.40.74100
UneditedVLM Backbone=Qwen3-VL-4B, Error Count=7002026.0200.180.480.430.73100