List your company

Code Generation on Energy-Aware Code Generation (Held-out problems)

63.5Train Success RateEnergy-SFT + EDP
33.44441.24749.0556.853Jul 6, 2026

Evaluation results

MethodLinks
Energy-SFT + EDPInitialization=E-SFT, Reward Function=EDP (Energy-Delay Product), Configuration Group=Multi-objective2026.0763.581.744.634.258.44.4512.63
Energy-SFT + energyInitialization=E-SFT, Reward Function=energy, Configuration Group=Pure single-axis vs. baseline2026.0753.48251.130.7133.44.4512.78
Energy-SFT + runtimeInitialization=E-SFT, Reward Function=runtime, Configuration Group=Cross-objective exploration2026.0753.282.251.130.5433.34.4514.44
AfterburnerInitialization=Runtime-SFT, Reward Function=runtime, Configuration Group=Pure single-axis vs. baseline2026.0749.388.564.623.3722.17.6412.19
Runtime-SFT + EDPInitialization=Runtime-SFT, Reward Function=EDP (Energy-Delay Product), Configuration Group=Multi-objective2026.0734.686.340.846.9351.57.6416.41