Predict Move on Predict Move 400 samples
98Prediction Move Legal %SFT8 XL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SFT8 XLSFT Train Tokens=60M, RL Samples=16k, Stage=RL2026.04 | 98 | 0.82 | 12 | |
| SFT8 + SFT9 XLcSFT Train Tokens=120M, RL Samples=16k, Stage=RL2026.04 | 98 | 0.83 | 22 | |
| gpt-oss-120b (Medium)SFT Train Tokens=–, RL Samples=–, Stage=SFT2026.04 | 94 | 0.66 | 2 | |
| SFT6 [BM]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 93 | 0.71 | 31 | |
| SFT9 XLSFT Train Tokens=60M, RL Samples=16k, Stage=RL2026.04 | 93 | 0.75 | 25 | |
| SFT8 [BM - All]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 90 | 0.71 | 14 | |
| SFT9 [BL - All]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 89 | 0.64 | 32 | |
| SFT7 [BL]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 85 | 0.63 | 64 | |
| SFT1 [RSPM]SFT Train Tokens=15M, RL Samples=8ka, Stage=RL2026.04 | 82 | 0.52 | 55 | |
| SFT2 [RSA]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 73 | 0.6 | 48 | |
| SFT4 [FBA]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 72 | 0.54 | 40 | |
| SFT5 [GS]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 71 | 0.54 | 42 | |
| SFT3 [VABP]SFT Train Tokens=15M, RL Samples=8k, Stage=RL2026.04 | 67 | 0.53 | 46 | |
| SFT8 [BM - All]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 60 | 0.6 | 3 | |
| SFT8 XLSFT Train Tokens=60M, RL Samples=16k, Stage=SFT2026.04 | 55 | 0.62 | 3 | |
| SFT8 + SFT9 XLcSFT Train Tokens=120M, RL Samples=16k, Stage=SFT2026.04 | 54 | 0.58 | 2 | |
| SFT9 XLSFT Train Tokens=60M, RL Samples=16k, Stage=SFT2026.04 | 52 | 0.6 | 3 | |
| SFT9 [BL - All]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 51 | 0.6 | 7 | |
| SFT7 [BL]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 48 | 0.62 | 2 | |
| SFT4 [FBA]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 44 | 0.59 | 3 | |
| SFT6 [BM]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 44 | 0.64 | 4 | |
| Llama 4 MaverickSFT Train Tokens=–, RL Samples=–, Stage=SFT2026.04 | 42 | 0.62 | 1 | |
| SFT3 [VABP]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 40 | 0.61 | 3 | |
| SFT2 [RSA]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 37 | 0.63 | 5 | |
| SFT5 [GS]SFT Train Tokens=15M, RL Samples=8k, Stage=SFT2026.04 | 36 | 0.6 | 3 | |
| SFT1 [RSPM]SFT Train Tokens=15M, RL Samples=8ka, Stage=SFT2026.04 | 34 | 0.63 | 4 | |
| Qwen2.5 7B-InstructSFT Train Tokens=–, RL Samples=–, Stage=SFT2026.04 | 8 | 0.56 | 6 |