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

Hallucination Evaluation on HallBench

73.6Accuracy

ToR-DAPO

Updated 6d ago

Evaluation Results

Method	Links
ToR-DAPO 2026.03		73.6
ToR-GRPO 2026.03		72.4
NoisyRollout-7B 2026.03		72.1
DR-MMSearchAgent (w/ SPAI) 2026.04		72
ToR-DAPO 2026.03		71.8
ThinkLite-7B-VL 2026.03		71
OpenVLThinker-7B 2026.03		70.8
ThinkLite-VL-7B 2026.04		70.2
VL-Rethinker-7B 2026.04		69.9
GRPO 2026.03		69.8
VLAA-Thinker-7B 2026.03		68.6
MM-Eureka-Qwen-7B 2026.03		68.3
DAPO 2026.03		67.9
InternVL-2.5-8B 2026.03		67.3
R1-OneVision-7B 2026.03		65.6
Qwen2.5-VL-7B 2026.04		65.2
Qwen2.5-VL-7B-Instruct 2026.03		64.6
Ovis2 2025.12		60.8
ToR-DAPO 2026.03		60.8
OpenVLThinker-7B 2026.04		59.1
ToR-DAPO 2026.03		56.3
Claude-3.7-Sonnet 2026.04		55.4
WeThink-7B 2026.04		55.1
GPT-4o 2026.04		55
DAPO 2026.03		54.7
Latent Denoising 2026.04		54.7
Baseline 2026.04		52.9
AndesVL 2025.12		51.8
SAIL-VL2 2025.12		51.7
HyperVL ViTL 2025.12		51.5
Qwen3-VL 2025.12		51.4
SAIL-VL1.5 2025.12		49.8
Vision-R1-LlamaV-CI-11B 2025.03		49.5
HyperVL 2025.12		49.2
Mulberry-Llama-11B 2025.03		48.9
InternVL3.5 2025.12		48.6
LLaVA-OneVision-7B 2026.03		48.4
LLaVA-Cot-11B 2025.03		47.8
Unicorn 2025.03		43.01
InternVL3 2025.12		42.5
Qwen2-VL 2025.12		41.7
Bunny 2025.03		41.26
Unicorn* 2025.03		40.38
Llama-3.2-11B-V 2025.03		40.3
LLaVA-1.5 2025.03		38.84
Latent Denoising 2026.04		30.7
Latent Denoising 2026.04		29.6
Baseline 2026.04		28.3
Baseline 2026.04		27.6