List your company

Counter-Perception Discrimination on CP-Bench (dev)

94.3F1 ScoreClaude-3-5-Sonnet
8.70830.92953.1575.371Apr 29, 2025

Evaluation results

MethodLinks
Claude-3-5-SonnetSource=Closed-sourced (API)2025.0494.3
GLM-4vSource=Closed-sourced (API)2025.0488.2
LLaVA-v1.5-13B + AntidoteStrategy=Antidote (Post-training)2025.0488
GPT-4v-0409Source=Closed-sourced (API)2025.0486
LLaVA-Next-Mistral-7B + AntidoteStrategy=Antidote (Post-training)2025.0484.4
GPT-4o-0806Source=Closed-sourced (API)2025.0484.2
LLaVA-v1.5-7B + AntidoteStrategy=Antidote (Post-training)2025.0482.9
GPT-4o-mini-0718Source=Closed-sourced (API)2025.0481.8
LLaVA-Next-Vicuna-13BSource=Open-sourced2025.0465.1
InternVL-2-ProSource=Closed-sourced (API)2025.0460.3
LLaVA-Next-Vicuna-7BSource=Open-sourced2025.0448.7
InternVL2-8BSource=Open-sourced2025.0447.1
LLaVA-Next-Mistral-7BCategory=Baseline2025.0443.6
Cogvlm2-19BSource=Open-sourced2025.0442.8
Qwen2-VL-72BSource=Open-sourced2025.0442.6
InternVL2-26BSource=Open-sourced2025.0441.2
Phi-3-VisionSource=Open-sourced2025.0438.8
MiniCPM-V2.5-8BSource=Open-sourced2025.0437
Qwen2-VL-7BSource=Open-sourced2025.0433.7
LLaVA-v1.5-7B + SeVaStrategy=SeVa2025.0425.4
InstructBLIP-7BSource=Open-sourced2025.0417.8
LLaVA-v1.5-7B + HA-DPOStrategy=HA-DPO2025.0413.1
LLaVA-v1.5-7BCategory=Baseline2025.0412.4
LLaVA-v1.5-7B + VCDStrategy=VCD2025.0412
LLaVA-v1.5-13BCategory=Baseline2025.0412