List your company

Visual Question Answering on iNaturalist (test)

46.6AccuracyGemma
3.12814.41425.736.986Jul 6, 2026

Evaluation results

MethodLinks
GemmaMitigation Protocol=After (Probe-guided)2026.0746.636.5
Qwen3Mitigation Protocol=After (Probe-guided)2026.0745.639.3
LlamaMitigation Protocol=After (Probe-guided)2026.0745.533.4
Qwen2.5Mitigation Protocol=After (Probe-guided)2026.0739.634.8
LlamaMitigation Protocol=Before2026.0712.1—
GemmaMitigation Protocol=Before2026.0710.1—
Qwen3Mitigation Protocol=Before2026.076.3—
Qwen2.5Mitigation Protocol=Before2026.074.8—