E-commerce Pricing Decision Alignment on E-commerce Pricing Expert Trajectories (Offline Evaluation)
2.836Avg. Q-ScoreAIGP
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| AIGPalignment_stage=SFT+DPO2026.06 | 2.836 | 0.062 | 82.51 | 11.83 | 11.86 | |
| AIGPalignment_stage=DPO-only2026.06 | 2.794 | 0.067 | 79.26 | 11.81 | 11.84 | |
| RL-DT2026.06 | 2.694 | 0.078 | 73.89 | 11.65 | 11.71 | |
| AIGPmode=FullSeq2026.06 | 2.591 | 0.082 | 71.32 | 11.73 | 11.78 | |
| Qwen3-30B-A3B2026.06 | 2.585 | 0.085 | 69.22 | 11.74 | 11.8 | |
| Price-Sales2026.06 | 2.564 | 0.084 | 71.08 | 11.62 | 11.68 | |
| Qwen3-235B-A22B2026.06 | 2.556 | 0.089 | 67.07 | 11.69 | 11.76 | |
| AIGPalignment_stage=SFT-only2026.06 | 2.554 | 0.086 | 68.98 | 11.71 | 11.77 | |
| RL-SAC2026.06 | 2.494 | 0.097 | 63.95 | 11.52 | 11.57 | |
| RL-DDPG2026.06 | 2.364 | 0.11 | 56.91 | 11.41 | 11.47 |