Visual Pattern Recognition on MMVP
80.9AccuracyQwen2.5VL 7B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2.5VL 7B# Vis tok.=14002025.04 | 80.9 | |
| FLARE-X 3B# Vis tok.=14002025.04 | 79.8 | |
| FLARE-X 8B# Vis tok.=14002025.04 | 79.7 | |
| InternVL3.5-14B +FINER-TuningBackbone=InternVL3.5-14B, Training Strategy=FINER-Tuning2026.03 | 78.7 | |
| FLARE-L 8B# Vis tok.=6302025.04 | 78.6 | |
| InternVL3.5-14BBackbone=InternVL3.5-14B2026.03 | 78.3 | |
| Qwen2.5-VL-7B +FINER-TuningBackbone=Qwen2.5-VL-7B, Training Strategy=FINER-Tuning2026.03 | 77.3 | |
| InternVL3.5-8B +FINER-TuningBackbone=InternVL3.5-8B, Training Strategy=FINER-Tuning2026.03 | 77 | |
| Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.03 | 76.7 | |
| InternVL3.5-8BBackbone=InternVL3.5-8B2026.03 | 76.7 | |
| Qwen2.5VL 3B# Vis tok.=14002025.04 | 76.7 | |
| FLARE-L 3B# Vis tok.=6302025.04 | 76.3 | |
| Florence-VL 8B# Vis tok.=5762025.04 | 73.3 | |
| Eagle 8B# Vis tok.=10242025.04 | 71.6 | |
| OmniLMM-12B +RLAIF-VBackbone=OmniLMM-12B, Training Strategy=RLAIF-V2026.03 | 70 | |
| OmniLMM-12BBackbone=OmniLMM-12B2026.03 | 69.7 | |
| LLaVA-1.6-7B +FINER-TuningBackbone=LLaVA-1.6-7B, Training Strategy=FINER-Tuning2026.03 | 68.7 | |
| Phi 3.5-Vision# Vis tok.=21002025.04 | 67.7 | |
| LLaVA-1.6-7BBackbone=LLaVA-1.6-7B2026.03 | 65 | |
| Florence-VL 3B# Vis tok.=5762025.04 | 64.7 | |
| Cambrain 8B# Vis tok.=5762025.04 | 51.3 | |
| FedAvg-TAsetting=MI-I2026.03 | 33.9 | |
| FedAvgsetting=MI-I2026.03 | 33.6 | |
| Fed-CMPsetting=MI-I2026.03 | 33 | |
| FedAdamsetting=MI-I2026.03 | 32.4 | |
| MOONsetting=MI-I2026.03 | 31.9 | |
| FedAvg-TIEssetting=MI-I2026.03 | 31.4 | |
| FedProxsetting=MI-I2026.03 | 30.3 | |
| Local Trainingsetting=MI-I2026.03 | 29.9 | |
| FedAvg-DAREsetting=MI-I2026.03 | 26.5 |