List your company

Text Generation on UltraRM

67.5Median AdvDPO-PoP-random
13.9427.84541.7555.655Sep 26, 2025

Evaluation results

MethodLinks
DPO-PoP-randomBackbone=Llama-3.2-3b2025.0967.562.4
DPO-margin-gtBackbone=Llama-3.2-3b2025.0958.661.3
DPO-PoP-iterBackbone=Llama-3.2-3b2025.0938.957.8
DPO-margin-1Backbone=Llama-3.2-3b2025.0922.754.9
DPO-margin-gt-scaledBackbone=Llama-3.2-3b2025.091653.7