List your company
ResearchFormerly WizWand

Visual Question Answering on ScienceQA (SQA)

89.3SQA AccuracyLocoRE
58.72466.66274.682.538Jan 28, 2026Feb 17, 2026Mar 10, 2026Mar 31, 2026Apr 21, 2026May 12, 2026Jun 2, 2026
Updated 5mo ago

Evaluation results

MethodLinks
LocoREBackbone=Qwen2.5-VL, Model Size=32B, Inference Strategy=LocoRE2026.0189.3
Qwen2.5-VL-32BBackbone=Qwen2.5-VL, Model Size=32B, Inference Strategy=Base2026.0189
SGRS + LocoREBackbone=Qwen2.5-VL, Model Size=7B, Inference Strategy=SGRS + LocoRE2026.0182.1
LocoREBackbone=Qwen2.5-VL, Model Size=7B, Inference Strategy=LocoRE2026.0180.8
Qwen2.5-VL-7BBackbone=Qwen2.5-VL, Model Size=7B, Inference Strategy=Base2026.0179
SGRS + LocoREBackbone=LLaVA-1.5, Model Size=13B, Inference Strategy=SGRS + LocoRE2026.0175.5
SGRS + LocoREBackbone=Qwen2-VL, Model Size=7B, Inference Strategy=SGRS + LocoRE2026.0175.3
VisionZipBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0674.4
LocoREBackbone=Qwen2-VL, Model Size=7B, Inference Strategy=LocoRE2026.0174.2
DARTBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0674.2
Qwen2-VL-7BBackbone=Qwen2-VL, Model Size=7B, Inference Strategy=Base2026.0174.1
VisionZipBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0674
DARTBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0673.8
Zoo-PruneBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0673.4
STSBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0673.2
DARTBackbone=LLaVA-1.5-13B, Token Budget=32 Tokens2026.0673.2
VisionZipBackbone=LLaVA-1.5-13B, Token Budget=32 Tokens2026.0672.9
VanillaBackbone=LLaVA-1.5-13B, Token Budget=Full Tokens2026.0672.8
DivPruneBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0672.8
AgilePruneBackbone=LLaVA-1.5-13B, Token Budget=128 Tokens2026.0672.8
STSBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0672.6
STSBackbone=LLaVA-1.5-13B, Token Budget=32 Tokens2026.0672.6
LocoREBackbone=LLaVA-1.5, Model Size=13B, Inference Strategy=LocoRE2026.0172.5
Zoo-PruneBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0672.1
AgilePruneBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0672
LLaVA-1.5-13BBackbone=LLaVA-1.5, Model Size=13B, Inference Strategy=Base2026.0171.6
DivPruneBackbone=LLaVA-1.5-13B, Token Budget=64 Tokens2026.0671.3
SGRS + LocoREBackbone=Intern-VL, Model Size=13B, Inference Strategy=SGRS + LocoRE2026.0171
DivPruneBackbone=LLaVA-1.5-13B, Token Budget=32 Tokens2026.0670.9
LocoREBackbone=Intern-VL, Model Size=13B, Inference Strategy=LocoRE2026.0170.4
Intern-VL-13BBackbone=Intern-VL, Model Size=13B, Inference Strategy=Base2026.0170.1
CAMDBase Model=LLaVA-1.52026.0368
SGRS + LocoREBackbone=Intern-VL, Model Size=7B, Inference Strategy=SGRS + LocoRE2026.0167.9
LLaVA-1.5-7B + SIDBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=SID2026.0167.8
SGRS + LocoREBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=SGRS + LocoRE2026.0167.8
LocoREBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=LocoRE2026.0167.5
LLaVA-1.5-7B + FarSightBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=FarSight2026.0167.4
FarSightBase Model=LLaVA-1.52026.0367.4
LLaVA-1.5-7B + VissinkBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=Vissink2026.0167
CAMDBase Model=Video-LLaVA2026.0366.9
LocoREBackbone=Intern-VL, Model Size=7B, Inference Strategy=LocoRE2026.0166.4
Intern-VL-7BBackbone=Intern-VL, Model Size=7B, Inference Strategy=Base2026.0166.2
FarSightBase Model=Video-LLaVA2026.0366.2
LLaVA-1.5-7B + TAMEBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=TAME2026.0166
LLaVA-1.5-7BBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=Base2026.0165.5
LLaVA-1.5-7B + OPERABackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=OPERA2026.0164.9
OPERABase Model=LLaVA-1.52026.0364.9
Video-LLaVADecoding Strategy=Greedy2026.0364.6
LLaVA-1.5Decoding Strategy=Greedy2026.0364.5
CGDBase Model=LLaVA-1.52026.0364
CAMDBase Model=InstructBLIP2026.0363.6
FarSightBase Model=InstructBLIP2026.0363.4
LLaVA-1.5-7B + VCDBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=VCD2026.0163.3
VCDBase Model=LLaVA-1.52026.0363.3
CAMDBase Model=Chat-UniVi2026.0363
LLaVA-1.5-7B + ICDBackbone=LLaVA-1.5, Model Size=7B, Inference Strategy=ICD2026.0162.8
ICDBase Model=LLaVA-1.52026.0362.8
FarSightBase Model=Chat-UniVi2026.0362.4
InstructBLIPDecoding Strategy=Greedy2026.0362.1
Chat-UniViDecoding Strategy=Greedy2026.0359.9