Vision-Language Modeling on MVision
50.5AccuracyZPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| ZPPOModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 50.5 | |
| GRPOModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 43.4 | |
| GRPOModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 43.1 | |
| On-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 42.7 | |
| On-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 41.9 | |
| Off-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 41.8 | |
| Off-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 40.4 | |
| Qwen3.5Model Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 38.4 | |
| ZPPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 29.2 | |
| GRPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 20.9 | |
| GRPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 20.1 | |
| On-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 19.4 | |
| On-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 18.7 | |
| Off-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 17.9 | |
| Off-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 17.3 | |
| Qwen3.5Model Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 16.4 |