List your company

Code Generation on HumanEval+ (Score)

34.76ScoreARES-RL
17.017621.623826.2330.8362May 22, 2026May 29, 2026Jun 5, 2026Jun 12, 2026Jun 19, 2026Jun 26, 2026Jul 3, 2026

Evaluation results

MethodLinks
ARES-RLBackbone=Qwen3-32B2026.0534.76
WebscaleBackbone=Qwen3-32B2026.0533.54
NaturalReasoningBackbone=Qwen3-32B2026.0532.93
CPTBackbone=Qwen3-32B2026.0532.32
ARES-SFTBackbone=Qwen3-32B2026.0531.1
Olmo3 BaseTuning Protocol=Freezing param.2026.0720.1
LMK token tuningTuning Protocol=Freezing param.2026.0719.5
Olmo3 512swaTuning Protocol=Full param. CPT2026.0719.5
HiLS-Attn HoPE-Q-CalTuning Protocol=Full param. CPT2026.0718.9
HiLS-Attn RoPE-Q-CalTuning Protocol=Full param. CPT2026.0718.9
HiLS-Attn NoPE-Q-CalTuning Protocol=Full param. CPT2026.0717.7