General Knowledge on MMLU-Pro 0-shot
44.4MMLU-Pro Score (0-shot)MARS-7B
Evaluation results
| Method | Links | |
|---|---|---|
| MARS-7BModel Scale=7B, Block Size (B)=4, One-token mode (τ=1.0)=true2026.04 | 44.4 | |
| AR SFTModel Scale=7B, One-token mode (τ=1.0)=true2026.04 | 43.9 | |
| MARS-0.5BModel Scale=0.5B, Block Size (B)=4, SFT Loss Configuration=True, One-token mode (τ=1.0)=true2026.04 | 12.4 | |
| MARS-0.5B w/o SFT lossModel Scale=0.5B, Block Size (B)=4, SFT Loss Configuration=False, One-token mode (τ=1.0)=true2026.04 | 12.3 | |
| AR SFT (5 ep)Model Scale=0.5B, Training Epochs=5, One-token mode (τ=1.0)=true2026.04 | 11.9 | |
| Base (no fine-tune)Model Scale=7B, One-token mode (τ=1.0)=true2026.04 | 10.5 | |
| AR SFT (10 ep)Model Scale=0.5B, Training Epochs=10, One-token mode (τ=1.0)=true2026.04 | 9.3 | |
| Base (no fine-tune)Model Scale=0.5B, One-token mode (τ=1.0)=true2026.04 | 2.4 | |
| Block DiffusionModel Scale=0.5B, Block Size (B)=4, One-token mode (τ=1.0)=true2026.04 | 2 |