List your company
ResearchFormerly WizWand

Visual Question Answering on VQAv2 (test-dev)

86.1AccuracyPaLI-X-55B
35.45248.60161.7574.899Oct 5, 2023Mar 4, 2024Aug 3, 2024Jan 2, 2025Jun 3, 2025Nov 2, 2025Apr 3, 2026
Updated 6mo ago

Evaluation results

MethodLinks
PaLI-X-55BModel Type=Specialist SOTA2023.1086.1
PaLI-XCategory=Specialist2023.1286
NVILA# Parameters=8B2024.1285.4
NVLM-D-1.0# Parameters=78B2024.1285.4
LLaVA-OneVision# Parameters=72B2024.1285.2
SMOLA-PALI-XFTCategory=Generalist, Fine-tuned=true2023.1285
NVILA-Lite# Parameters=8B2024.1285
NVILA# Parameters=15B2024.1284.8
CogVLMCategory=Specialist2023.1284.7
Florence-VL# Parameters=8B2024.1284.7
PaLI-XFTCategory=Generalist, Fine-tuned=true2023.1284.4
VILA-1.5# Parameters=40B2024.1284.3
BEIT-3Category=Specialist2023.1284.2
LLaVA-OneVision# Parameters=8B2024.1284
Cambrian-1# Parameters=34B2024.1283.8
NVILA-Lite# Parameters=15B2024.1283.7
LLaVA-NeXT# Parameters=34B2024.1283.7
SMOLA48-PaLI-3FTCategory=Generalist, Fine-tuned=true2023.1283.6
CogVLMCategory=Generalist2023.1283.4
VILA-1.5# Parameters=8B2024.1283
Qwen2-VL# Parameters=8B2024.1282.9
VILA-1.5# Parameters=13B2024.1282.8
PaLI-3FTCategory=Generalist, Fine-tuned=true2023.1282.5
LLaVA-1.5-HDLLM=Vicuna-13B, Image Size=448^2, Pretrain Sample Size=558K, Finetune Sample Size=665K2023.1081.8
GIT2Category=Specialist2023.1281.7
InternVL-ChatLLM=Vicuna-13B, Res.=336, glue layer=QLLAMA2023.1281.2
Cambrian-1# Parameters=8B2024.1281.2
Cambrian-1# Parameters=13B2024.1281.2
LOVA³-7BLLM=Vicuna-7B2024.0580.3
InternVL-ChatLLM=Vicuna-13B, Res.=336, glue layer=MLP2023.1280.2
Gemini 1.5 Pro2024.1280.2
Pixtral# Parameters=12B2024.1280.2
LLaVA-1.5LLM=Vicuna-13B, Res.=3362023.1280
PaLM-ECategory=Specialist2023.1280
LLaVA-1.5LLM=Vicuna-13B, Image Size=336^2, Pretrain Sample Size=558K, Finetune Sample Size=665K2023.1080
LLaVA-1.5# Parameters=13B2024.1280
LLaVA 1.52023.1280
RLHF-VBackbone=Vicuna v1.0 13B, Visual Module=BEiT-3, Resolution=4482023.1280
Qwen-VLCategory=Generalist2023.1279.5
Qwen-VL-Chat2023.1279.5
InternVL-ChatLLM=Vicuna-7B, Res.=336, glue layer=MLP2023.1279.3
Qwen-VLLLM=Qwen-7B2023.1278.8
Qwen-VLLLM=Qwen-7B, Image Size=448^2, Pretrain Sample Size=1.4B+, Finetune Sample Size=50M+2023.1078.8
GPT-4o2024.1278.7
LLaVA-1.5LLM=Vicuna-7B, Res.=3362023.1278.5
LLaVA-1.5LLM=Vicuna-7B, Image Size=336^2, Pretrain Sample Size=558K, Finetune Sample Size=665K2023.1078.5
LLaVA-1.5LLM=Vicuna-7B2024.0578.5
LLaVA-1.5# Parameters=7B2024.1278.5
Qwen-VL-ChatLLM=Qwen-7B2023.1278.2
Qwen-VL-ChatLLM=Qwen-7B, Image Size=448^2, Pretrain Sample Size=1.4B+, Finetune Sample Size=50M+2023.1078.2
Unified-IOCategory=Generalist2023.1277.9
ShikraLLM=Vicuna-13B2023.1277.4
ShikraLLM=Vicuna-13B, Image Size=224^2, Pretrain Sample Size=600K, Finetune Sample Size=5.5M2023.1077.4
UncompressedModel=BLIP, Pruning Mode=/, Pruning Ratio=/, GFLOPs=186.12026.0477.4
GPT-4VZero-shot=true2023.1277.2
InternVL2# Parameters=8B2024.1276.7
CoMPModel=BLIP, Pruning Mode=C, Pruning Ratio=0.7, GFLOPs=59.72026.0476.5
MADTPModel=BLIP, Pruning Mode=T, Pruning Ratio=0.7, GFLOPs=61.62026.0476.3
LOVA³-1.5BLLM=Phi-1.5B2024.0575.8
UPopModel=BLIP, Pruning Mode=P, Pruning Ratio=0.7, GFLOPs=62.32026.0474.5
LLaVA-Phi-1.5LLM=Phi-1.5B2024.0573.2
InternVL-ChatLLM=Vicuna-7B, Res.=224, glue layer=QLLAMA2023.1272.3
InternVL-ChatLLM=Vicuna-13B, Res.=224, glue layer=QLLAMA2023.1271.7
Claude 3.5 Sonnet2024.1270.7
MH-MoEExperts Number=82024.0470.1
X-MoEExperts Number=82024.0468.4
Dense2024.0465.9
IDEFICS-80BLLM=LLaMA-65B2023.1265
BLIP-2LLM=Vicuna-13B, Image Size=224^2, Pretrain Sample Size=129M, Finetune Sample Size=None2023.1065
IDEFICS-80BLLM=LLaMA-65B, Image Size=224^2, Pretrain Sample Size=353M, Finetune Sample Size=1M2023.1060
BLIP-2LLM=Vicuna-7b, Data=104M, steps=330k, Zero-shot=true2023.1057.8
EVLGenLLM=Vicuna-7b, Data=104M, steps=150k, Zero-shot=true2023.1055.5
EVLGenLLM=Vicuna-7b, Data=104M, steps=250k, Zero-shot=true2023.1054.8
EVLGenLLM=Vicuna-7b, Data=11M, steps=150k, Zero-shot=true2023.1054.6
EVLGenLLM=Vicuna-7b, Data=104M, steps=90k, Zero-shot=true2023.1053.4
IDEFICS-9BLLM=LLaMA-7B, Image Size=224^2, Pretrain Sample Size=353M, Finetune Sample Size=1M2023.1050.9
InstructBLIPLLM=Vicuna-13B2023.1249.5
InstructBLIPLLM=Vicuna-7B2023.1249.2
BLIP-2LLM=Vicuna-13B2023.1241
IDEFICS-80B-ILLM=LLaMA-65B2023.1237.4