List your company

Hallucination Evaluation on MME Hallucination

195Existence ScoreDAC
140.4154.575168.75182.925May 28, 2024Sep 26, 2024Jan 26, 2025May 28, 2025Sep 27, 2025Jan 27, 2026May 29, 2026

Evaluation results

MethodLinks
DACDecoding strategy=greedy decoding2025.02195158.33133.33170656.67
AGLABackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04195153.9129.4161.7640
ACTBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04195153.3150170668.3
TAMEBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04193137.3139164.7634
ONLYBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04191.7145.6136.7161.7635.6
CMACBackbone=LLaVA1.5, Decoding=Ours2025.01191.25135.41123.33162.5612.49
CoFi-DecBackbone=LLaVA-1.52025.12190.26144.43133.71165.62627.39
SIDDecoding strategy=greedy decoding2025.02190148.33128.33175641.66
CCADecoding strategy=greedy decoding2025.02190148.33128.33175641.66
HALCBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04190143.3128.3160621.6
CCABackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04190148.3128.3153641.7
EAHBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04190108.3145160.7604
VARBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04190148.3138.3155631.3
MEMVRBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04190155133.3170.6648.3
OPERABackbone=LLaVA-1.5-7B2026.05190133.33121.67155600
EVASBackbone=LLaVA-1.5-7B2026.05190140121.67160.33612
TAMEBackbone=LLaVA-1.5-7B2026.05190133.66128.33163614.99
AvISCModel=LLaVA 1.52024.05189.29104.76106.19127.86528.09
DeFGBackbone=LLaVA-1.52025.12188.33142.5131.33163.17625.83
RITUALBackbone=LLaVA-1.52025.12187.53139.58125163.33626.29
WoodpeckerBackbone=LLaVA-1.52025.12187.53126.25126.66149.17589.58
RITUALBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04187.5139.6125164.2616.3
VCDBackbone=LLaVA-1.52025.12186.67125.56128.89139.45580.56
M3IDBackbone=LLaVA-1.52025.12186.67128.33131.63151.67597.5
VCDBackbone=LLaVA1.5, Decoding=VCD2025.01186.25116.67119.17159.17581.25
DOPRABackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04185.7138.3120.7133577.7
ICDBackbone=LLaVA1.5, Decoding=ICD2025.01185117.91117.5162.08582.5
VCDBackbone=LLaVA-1.5-7B2026.05185126.67121.67135568.33
FuzzyCDBackbone=LLaVA-1.5-7B2026.05185148.33123.33153.33609.99
YARDBackbone=LLaVA-1.5-7B2026.05185140125163.33613.33
AvISCModel=InstructBLIP2024.05184.7682.8574.76131.43473.8
VCDBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04184.7137.3128.7153603.7
VCDDecoding strategy=greedy decoding2025.02184.66138.33128.67153604.66
SIDBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04183.9132.2127.8155.9599.8
BaselineBackbone=LLaVA1.5, Decoding=Baseline2025.01183.75110115.83157.08566.67
CMACBackbone=InstructBLIP, Decoding=Ours2025.01183.758568.75131.67469.17
OPERABackbone=LLaVA-1.52025.12183.33137.22122.78155598.33
HALCBackbone=LLaVA-1.52025.12183.33133.33109.58155581.24
OPERABackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04180.7133.3111.7123.3549
OPERADecoding strategy=beam search2025.02180.67133.33123.33155592.33
DolaBackbone=LLaVA-1.5-7B, Evaluation Protocol=zero-shot2026.04180.1127.4119.3154.6594.1
LLaVA-1.5-7BBackbone=LLaVA-1.5-7B2026.05180110121.33141.67553
ICDBackbone=LLaVA-1.5-7B2026.05180126.67113.33143.33563.33
AVISCBackbone=LLaVA-1.5-7B2026.05180138.33123.33158.33599.99
M3IDModel=LLaVA 1.52024.05178.33107.2296.39127.5509.44
DoLaBackbone=LLaVA-1.52025.12176.67113.3390.55141.67522.22
BaselineDecoding strategy=greedy decoding2025.02175.67124.67114151565.33
M3IDModel=InstructBLIP2024.05173.8989.7272.72110.56446.88
RegularBackbone=LLaVA-1.52025.12173.75121.67117.92149.17562.5
Base DecodingModel=LLaVA 1.52024.05173.57110100.47125.24509.28
M3IDBackbone=LLaVA-1.5-7B2026.05173.33106.6796.67155531.67
VCDModel=InstructBLIP2024.05172.6298.3371.9117.14459.99
VCDModel=LLaVA 1.52024.05172.14117.14103.33119.52512.14
Base DecodingModel=InstructBLIP2024.05170.1989.5267.62114.76442.09
VCDBackbone=InstructBLIP, Decoding=VCD2025.01166.6782.0867.08130445.83
ICDBackbone=InstructBLIP, Decoding=ICD2025.01166.6780.8367.92128.33443.75
CMACBackbone=Qwen-VL, Decoding=Ours2025.01157.75118.75132.83168.75578.08
BaselineBackbone=InstructBLIP, Decoding=Baseline2025.01154.1786.6758.33123.75422.91
VCDBackbone=Qwen-VL, Decoding=VCD2025.01148.58113.67133164.92560.17
ICDBackbone=Qwen-VL, Decoding=ICD2025.01144.75116.44135.17162.75559.11
BaselineBackbone=Qwen-VL, Decoding=Baseline2025.01142.5116.67134.08157.08550.33