Our new X account is live! Follow @wizwand_team for updates

Hallucination Evaluation on POPE

94.42Accuracy

MIRROR(ours)

Updated 2d ago

Evaluation Results

Method
MIRROR(ours) 2026.02	94.42	-	-	-	-	-	-
SPHINX-1k 2023.11	90.8	-	-	-	-	-	-
InternVL3 2026.02	90.37	-	-	-	-	-	-
InternVL2.5-MPO 2025.01	89.8	-	-	-	-	-	-
Phantom-1.8B 2024.12	89.6	-	-	-	-	-	-
InternVL3 2026.02	89.6	-	-	-	-	-	-
LIVR 2026.02	89.5	-	-	-	-	-	-
DeepSeekVL-2 2025.01	89.4	-	-	-	-	-	-
Qwen3-VL-4B 2026.02	89.3	-	-	-	-	-	-
LLaVA-UHD 2025.01	89.1	-	-	-	-	-	-
InternVL2.5-MPO 2025.01	89.1	-	-	-	-	-	-
Visual Prompt 2024.08	88.9	-	-	-	-	-	-
Qwen2-VL-7B 2024.12	88.9	-	-	-	-	-	-
InternVL-3.5-4B 2026.02	88.9	-	-	-	-	-	-
DeepSeekVL-2 2025.01	88.8	-	-	-	-	-	-
CrystaL 2026.02	88.7	-	-	-	-	-	-
TroL-1.8B 2024.12	88.6	-	-	-	-	-	-
VLsI-7B 2024.12	88.6	-	-	-	-	-	-
SEA-PRIME 2024.08	88.4	-	-	-	-	-	-
VOLCANO 2023.11	88.3	-	87.7	-	-	-	-
PIIP-LLaVA 2025.01	88.2	-	-	-	-	-	-
VOLCANO 2023.11	88.2	-	87.7	-	-	-	-
LLaVA-HR-X 2024.03	88	-	-	-	-	-	-
MIRROR(w/o tool) 2026.02	87.95	-	-	-	-	-	-
PIIP-LLaVA 2025.01	87.9	-	-	-	-	-	-
PIIP-LLaVA 2025.01	87.9	-	-	-	-	-	-
VLSI-2B 2024.12	87.9	-	-	-	-	-	-
LLAVA-HR 2025.01	87.8	-	-	-	-	-	-
SEA-PRIME 2024.08	87.8	-	-	-	-	-	-
Qwen2-VL-2B 2024.12	87.8	-	-	-	-	-	-
MM1-MoE-7B×32 2024.12	87.8	-	-	-	-	-	-
TroL-7B 2024.12	87.8	-	-	-	-	-	-
LLaVA-HR 2024.03	87.8	-	-	-	-	-	-
Phantom-7B 2024.12	87.7	-	-	-	-	-	-
LLaVA-HR 2025.01	87.6	-	-	-	-	-	-
PIIP-LLaVA 2025.01	87.6	-	-	-	-	-	-
LLaVA-1.5* 2024.08	87.6	-	-	-	-	-	-
MM1-MoE-3B×64 2024.12	87.6	-	-	-	-	-	-
DeepSeek-VL-1.3B 2024.12	87.6	-	-	-	-	-	-
LLaVA-HR 2024.03	87.6	-	-	-	-	-	-
Qwen2-VL 2025.01	87.6	-	-	-	-	-	-
PIIP-LLaVA 2025.01	87.5	-	-	-	-	-	-
Mistral Instruct v0.1 7B 2025.12	87.5	-	-	-	-	-	-
VIG training + VAR 2026.02	87.5	-	86.99	-	-	-	-
VIG training + PAI 2026.02	87.49	-	86.12	-	-	-	-
VIG training 2026.02	87.47	-	85.93	-	-	-	-
AVG-LLaVA 2024.09	87.4	-	-	-	-	-	-
S2-Wrapper* 2024.08	87.4	-	-	-	-	-	-
SEA-PRIME 2024.08	87.4	-	-	-	-	-	-
MM1-3B 2024.12	87.4	-	-	-	-	-	-
VIG training + LACING 2026.02	87.39	-	86.19	-	-	-	-
LLaVA-1.5 2024.09	87.3	-	-	-	-	-	-
Moxin-7B 2025.12	87.3	-	-	-	-	-	-
SKILA 2026.02	87.3	-	-	-	-	-	-
SPHINX-2k 2024.09	87.2	-	-	-	-	-	-
LLaVA-NeXT-M3 2024.09	87.2	-	-	-	-	-	-
SPHINX-2k 2023.11	87.2	-	-	-	-	-	-
VAR 2026.02	87.12	-	86.1	-	-	-	-
VIG training + VCD 2026.02	87.12	-	86.11	-	-	-	-
Phantom-3.8B 2024.12	87.1	-	-	-	-	-	-
Mistral v0.1 7B 2025.12	87.1	-	-	-	-	-	-
LLaVA-1.5 7B 2026.02	87.08	-	85.9	-	-	-	-
SEA-PRIME 2024.08	87	-	-	-	-	-	-
MoE-LLaVA-1.8B×4 2024.12	87	-	-	-	-	-	-
SPHINX-intern2 2025.01	86.9	-	-	-	-	-	-
MG-LLaVA 2024.09	86.9	-	-	-	-	-	-
Llama-2 Chat 7B 2025.12	86.8	-	-	-	-	-	-
Llama-2 7B 2025.12	86.74	-	-	-	-	-	-
SEA-PRIME 2024.08	86.7	-	-	-	-	-	-
LLaVA-NeXT-13B 2024.12	86.7	-	-	-	-	-	-
VILA2-8B 2024.12	86.7	-	-	-	-	-	-
SAIL-VL 2025.01	86.7	-	-	-	-	-	-
LACING 2026.02	86.68	-	85.58	-	-	-	-
VCD 2026.02	86.61	-	86.53	-	-	-	-
MM1 2025.01	86.6	-	-	-	-	-	-
MM1-7B 2024.12	86.6	-	-	-	-	-	-
LLaVa v1.5 7B (Base) 2025.12	86.57	-	-	-	-	-	-
LLaVA-NeXT 2025.01	86.5	-	-	-	-	-	-
LLaVA-NeXT 2024.09	86.5	-	-	-	-	-	-
LLaVA-1.5* 2024.08	86.5	-	-	-	-	-	-
TroL-3.8B 2024.12	86.5	-	-	-	-	-	-
LLaVA-NeXT-7B 2024.12	86.5	-	-	-	-	-	-
Qwen2-VL 2025.01	86.5	-	-	-	-	-	-
Qwen2.5-VL-7B 2026.02	86.5	-	-	-	-	-	-
Qwen2.5-VL-7B 2026.02	86.45	-	-	-	-	-	-
InternVL-Chat 2024.09	86.4	-	-	-	-	-	-
TinyLLaVA 2024.08	86.4	-	-	-	-	-	-
DenseMLLM-4B 2026.02	86.4	-	-	-	-	-	-
VILA1.5-13B 2024.12	86.3	-	-	-	-	-	-
Qwen2.5-VL-3B 2026.02	86.21	-	-	-	-	-	-
AlignGPT 2024.08	86.2	-	-	-	-	-	-
LLaVA-1.5 2023.11	86.2	-	85.2	-	-	-	-
LLaVA-1.5 2023.11	86.1	-	85.1	-	-	-	-
OtterHD-8B 2024.09	86	-	-	-	-	-	-
LLaMA-VID 2024.08	86	-	-	-	-	-	-
LLaMA-VID 2024.08	86	-	-	-	-	-	-
AlignGPT 2024.08	86	-	-	-	-	-	-
LLaVA-1.5 2025.01	85.9	-	-	-	-	-	-
LLaVA-1.5 2025.01	85.9	-	-	-	-	-	-
LLaVA1.5-7B 2023.11	85.9	-	-	-	-	-	-

Showing 100 of 257 rows