Vision-Language Modeling on CharXiv
73.9AccuracyZPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| ZPPOModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 73.9 | |
| GRPOModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 73 | |
| GRPOModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 72.8 | |
| On-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 71.7 | |
| On-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 71.7 | |
| Qwen3.5Model Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 71.6 | |
| Off-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=true2026.06 | 71.4 | |
| Off-DistillModel Scale=2B, Prompt Replay Buffer Augmentation=false2026.06 | 71 | |
| ZPPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 63.9 | |
| GRPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 59.9 | |
| GRPOModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 57 | |
| On-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 56.4 | |
| Off-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=true2026.06 | 56 | |
| On-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 56 | |
| Off-DistillModel Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 55.1 | |
| Qwen3.5Model Scale=0.8B, Prompt Replay Buffer Augmentation=false2026.06 | 54.3 |