Share your thoughts, 1 month free Claude Pro on usSee more

Multimodal Instruction Following on LLaVA-Bench In-the-Wild

93.1Score

GPT-4V

Updated 1mo ago

Evaluation Results

Method	Links
GPT-4V 2024.01		93.1
InternLM-XComposer2 2024.01		81.8
Gemini-Pro 2024.01		79.9
Lyrics 2023.12		76.9
CSR 2024.05		74.7
CogVLM 2024.01		73.9
Qwen-VL-Plus 2024.01		73.7
ShareGPT4V 2023.12		72.6
CSR 2024.05		71.1
LLaVA-1.5 2023.12		70.7
LLaVA-1.5-13B 2024.05		70.7
SemDeDup 2026.04		70
POVID 2024.05		68.7
Perplexity 2026.04		68.3
Full 2026.04		67.9
Qwen-VL-Chat 2023.12		67.7
COINCIDE 2026.04		67.3
CLIP-Score 2026.04		66.2
ICONS 2026.04		66.1
DOSE 2026.04		65.8
Self-rewarding 2024.05		65.6
RLHF-V 2024.05		65.4
Random 2026.04		65
EL2N 2026.04		64.9
Self-Filter 2026.04		64.9
D2-Pruning 2026.04		63.9
Human-Prefer 2024.05		63.7
LLaVA-1.5-7B 2024.05		63.4
Self-Sup 2026.04		63.3
LLaVA 2023.12		63
Vlfeedback 2024.05		62.1
Self-rewarding 2024.05		61.2
InstructBLIP 2023.12		60.9
IDEFICS-80B 2023.12		56.9
BLIP-2 2023.12		38.1