List your company

Visual Hallucination Evaluation on AMBER Generative

6.3CHAIRiVicuna-7B
4.8445.2225.65.978Jun 22, 2026

Evaluation results

MethodLinks
Vicuna-7BBackbone=Vicuna-7B2026.066.352.230.83.3
HeRABackbone=Vicuna-7B2026.066.152.631.93.5
Vicuna-13BBackbone=Vicuna-13B2026.066.152.528.43.3
HeRABackbone=Vicuna-13B2026.066.152.930.33.2
Qwen2.5-14BBackbone=Qwen2.5-14B2026.066.152.329.43.6
Qwen2.5-3BBackbone=Qwen2.5-3B2026.065.85230.13.3
Qwen3-8BBackbone=Qwen3-8B2026.065.852.828.93.2
HeRABackbone=LLama3-8B2026.065.651.726.62.7
Qwen3-4BBackbone=Qwen3-4B2026.065.55328.23
HeRABackbone=Qwen3-4B2026.065.552.527.62.8
LLama3-8BBackbone=LLama3-8B2026.065.551.527.52.9
Qwen3-14BBackbone=Qwen3-14B2026.065.553.128.83.2
HeRABackbone=Qwen2.5-14B2026.065.451.9283.1
Qwen2.5-7BBackbone=Qwen2.5-7B2026.065.352.326.72.7
HeRABackbone=Qwen2.5-3B2026.065.252.327.62.8
HeRABackbone=Qwen3-8B2026.065.253.127.93
HeRABackbone=Qwen3-14B2026.0655326.73
HeRABackbone=Qwen2.5-7B2026.064.952.525.12.7