List your company
ResearchFormerly WizWand

Visual Question Answering on VQA v2 (Overall Score)

89.4Overall AccuracyDDP
56.53665.06873.682.132Nov 26, 2024Feb 23, 2025May 24, 2025Aug 21, 2025Nov 19, 2025Feb 16, 2026May 17, 2026
Updated 5mo ago

Evaluation results

MethodLinks
DDPbackbone=Gemini-3-Pro2026.0489.4
Gemini-3-pro2026.0485.6
GPT-4o2026.0482.3
LLaVA-CKDLLM=Qwen2.5-1.5B, #TS=1.2M2026.0580.04
baselineModel=LLaVA-v1.5-13B2025.0780
SENTINELModel=LLaVA-v1.5-13B2025.0779.9
ImpLLM=Qwen1.5-1.8B, #TS=1.5M2026.0579.2
vanilla-DPOModel=LLaVA-v1.5-13B2025.0779.2
Qwen-VL2026.0478.8
CompoDistillLLM=Qwen1.5-1.8B, #TS=1.2M2026.0578.8
LLaVA-v1.52026.0478.5
baselineModel=LLaVA-v1.5-7B2025.0778.5
VCDModel=LLaVA-v1.5-13B2025.0778.5
TPCL_Fix↑backbone=LXMERT2024.1178.42
SENTINELModel=LLaVA-v1.5-7B2025.0778.4
HSA-DPOModel=LLaVA-v1.5-13B2025.0778.3
OPERAModel=LLaVA-v1.5-7B2025.0778.2
Individuals2026.0578.11
EFUFModel=LLaVA-v1.5-7B2025.0778.1
FREE-Merging2026.0578.05
TPCL_Dyn↑backbone=LXMERT2024.1178.03
DiDi-Merging2026.0578.03
MMER2026.0577.95
HA-DPOModel=LLaVA-v1.5-7B2025.0777.6
POVIDModel=LLaVA-v1.5-7B2025.0777.2
VCDModel=LLaVA-v1.5-7B2025.0777
LLaVA-CKDLLM=Qwen2.5-0.5B, #TS=1.2M2026.0576.8
BunnyLLM=Qwen1.5-1.8B, #TS=2.6M2026.0576.6
DoLaModel=LLaVA-v1.5-7B2025.0776.3
MoE-LLaVALLM=Qwen1.5-1.8B, #TS=2.1M2026.0576.2
LVIDALLM=Llama-3.2-1B, #TS=1.2M2026.0576.2
TPOModel=LLaVA-v1.5-7B2025.0775.9
RLAIF-VModel=LLaVA-v1.5-7B2025.0775.2
SIMPLEAUGbackbone=LXMERT2024.1174.98
Sphinx-TinyLLM=TinyLlama-1.1B, #TS=15M2026.0574.7
MoVE-KD-v1.1LLM=MobLLaMA-1.4B, #TS=1.2M2026.0573.8
Gemini-1.5-pro2026.0473.2
LXMERT2024.1173.06
DGGbackbone=UpDn2024.1165.54
UpDn2024.1163.48
PWVQAbackbone=UpDn2024.1162.63
GGDbackbone=UpDn2024.1162.15
CVIVbackbone=UpDn2024.1161.93
NaiveMC2026.0559.73
EM-KDLLM=Qwen2.5-0.5B, #TS=1.3M2026.0557.8