AI Paper Atlas

Research map / Reinforcement learning / RL from feedback

Preference Optimization via Contrastive Divergence: Your Policy Is Secretly an NLL Estimator

Zhuotong Chen, Fang Liu, Xuan Zhu, Haozhu Wang, Jiayu Li, Yanjun Qi, Mohammad Ghavamzadeh

AAAI 2026 · Reinforcement learning › RL from feedback · approach: Reward modeling versus direct optimization · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use MC-kernel rejected sampling instead of random sampling (MC-PO vs DPO)DPO · AlpacaEval 2 / Arena-Hard, Mistral-7B-SFTwin-rate: 25.07 -> 30.86; 42.01 -> 52.75Improved
use MC-kernel rejected sampling instead of random samplingDPO · AlpacaEval 2 / Arena-Hard, Llama-3.1-8B-SFTwin-rate: 33.74 -> 35.84; 60.25 -> 63.77Improved
use MC-kernel rejected sampling in instruct setupDPO · AlpacaEval 2 / Arena-Hard, Llama-3.1-8B-Instructwin-rate: 64.22 -> 66.90; 75.88 -> 76.71Mixed
add noisy near-duplicate sample to candidates: MC-PO vs DPO (ablation)DPO · Noisy candidates, Llama-3.1-8Bwin-rate: 23.62 -> 28.98 (Alpaca); 50.51 -> 58.09 (Arena)Improved
select the noisy sample as rejected response (DPO(-)) (ablation)DPO · Noisy candidates, Llama-3.1-8Bwin-rate: 23.62 -> 1.08 (Alpaca)Hurt
increase rejected samples M from 1 to 3 with MC kernel (ablation)MC-PO · AlpacaEval 2, Llama-3.1-8B-SFTwin-rate: 35.84 -> 37.40Improved
increase rejected samples M from 1 to 3 with random sampling (ablation)DPO · AlpacaEval 2, Llama-3.1-8B-SFTwin-rate: 33.74 -> 34.36No clear effect
replace DPO 4x augmented pairs with MC-PO on original data (ablation)DPO · Llama-3.1-8B-SFTwin-rate: 34.18 -> 35.84 (Alpaca); 59.62 -> 63.77 (Arena)Improved
use Min sampling instead of MC kernel (ablation)MC-PO · Llama-3.1-8B-SFT Figure 2win-rateHurt
remove high-ranked candidates from the candidate set (ablation)MC-PO · Llama-3.1-8B-SFT Nectar x-negswin-rateHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy