List your company

LLM Alignment on HH-RLHF (test)

0.87DiversityRE-Control
0.559040.639770.72050.80123Jun 10, 2024Oct 2, 2024Jan 24, 2025May 18, 2025Sep 9, 2025Jan 1, 2026Apr 26, 2026

Evaluation results

MethodLinks
RE-ControlModel=Hermes3-8B2026.040.870.55-4.321—85.58473.1
Pref-CTRL (Margin + Regularizer)Model=Hermes3-8B2026.040.870.56-4.268—85.784.373.6
Pref-CTRL (Margin)Model=Hermes3-8B2026.040.860.55-4.301—86.783.773.2
DPOModel=Hermes3-8B2026.040.850.55-4.297—86.878.775.8
RE-CONTROL + PromptingBackbone=Vicuna-7B, Inference time (hour)=0.932024.060.830.5776.26780.3———
RE-CONTROLBackbone=Vicuna-7B, Inference time (hour)=0.852024.060.8240.5796.21475.6———
Static REBackbone=Vicuna-7B, Inference time (hour)=0.652024.060.8180.5685.90764.3———
PromptingBackbone=Vicuna-7B, Inference time (hour)=0.692024.060.8170.575.91366———
BaseBackbone=Vicuna-7B, Inference time (hour)=0.602024.060.8160.5685.89457.6———
CD prefix + PromptingBackbone=Vicuna-7B, Inference time (hour)=47.162024.060.8120.5936.1274.3———
CDBackbone=Vicuna-7B, Inference time (hour)=47.432024.060.8060.6085.45872.3———
CD prefixBackbone=Vicuna-7B, Inference time (hour)=32.132024.060.8050.5766.10574.6———
Pref-CTRL (Margin)Model=Vicuna-7B2026.040.750.53-5.358—80.782.572.4
PromptingBackbone=Falcon-7B, Inference time (hour)=0.592024.060.7460.624.0152.3———
RE-CONTROL + PromptingBackbone=Falcon-7B, Inference time (hour)=2.002024.060.7410.6194.08362.6———
RE-ControlModel=Vicuna-7B2026.040.720.54-5.408—81.985.473.3
DPOModel=Vicuna-7B2026.040.710.55-5.591—85.18981.2
BaseBackbone=Falcon-7B, Inference time (hour)=0.672024.060.7050.6133.43942.3———
Pref-CTRL (Margin + Regularizer)Model=Vicuna-7B2026.040.70.55-5.288—82.985.674.6
RE-CONTROLBackbone=Falcon-7B, Inference time (hour)=1.932024.060.6990.6153.51258———
Static REBackbone=Falcon-7B, Inference time (hour)=0.562024.060.6980.613.44952.6———
CD prefixBackbone=Falcon-7B, Inference time (hour)=48.132024.060.6480.5754.39749.6———
CD prefix + PromptingBackbone=Falcon-7B, Inference time (hour)=47.872024.060.5710.6383.61951.6———
MAVISBase Model=Llama-7B, Baseline Compared Against=MOD2025.08———56.5———
MAVISBase Model=Llama-7B, Baseline Compared Against=RSoup2025.08———67.5———
MAVISBase Model=Llama-7B, Baseline Compared Against=MOODPO2025.08———35.5———
MAVISBase Model=Llama-13B, Baseline Compared Against=MOD2025.08———59.5———
MAVISBase Model=Llama-13B, Baseline Compared Against=RSoup2025.08———62———
MAVISBase Model=Llama-13B, Baseline Compared Against=MOODPO2025.08———62———