Meme Classification on MMSD 2.0 (test)
86.9AccuracyGemini-3-Flash
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini-3-FlashMode=zero-shot2026.06 | 86.9 | 86.2 | |
| Claude-Sonnet-4.6Mode=zero-shot2026.06 | 85.2 | 84.6 | |
| Intent ProjectionBackbone=InternVL3-8B2026.06 | 84.8 | 84 | |
| Intent ProjectionBackbone=Qwen3-VL-8B2026.06 | 83.7 | 82.9 | |
| InternVL3-8BSFT=true, Chain=true2026.06 | 82.5 | 81.7 | |
| GPT-5.4-miniMode=zero-shot2026.06 | 82.4 | 81.8 | |
| Qwen3-VL-8BSFT=true, Chain=true2026.06 | 81.2 | 80.4 | |
| Intent ProjectionBackbone=LLaVA-OneVision-7B2026.06 | 80.9 | 80.1 | |
| LLaVA-OneVision-7BSFT=true, Chain=true2026.06 | 78.4 | 77.6 | |
| InternVL3-8BMode=base2026.06 | 76.1 | 75.3 | |
| Qwen3-VL-8BMode=base2026.06 | 74.8 | 74 | |
| LLaVA-OneVision-7BMode=base2026.06 | 71.3 | 70.5 |