Conditional Equivalence of DPO and RLHF: Implicit Assumption, Failure Modes, and Provable Alignment
The paper discusses the conditional equivalence of Direct Preference Optimization (DPO) and Reinforcement Learning from Human Feedback (RLHF). It highlights that the theoretical equivalence relies on an implicit assumption that is often violated in practice. The authors propose Constrained Preference Optimization (CPO) as a solution to ensure provable alignment while maintaining simplicity.
- ▪Direct Preference Optimization (DPO) is presented as a simpler alternative to Reinforcement Learning from Human Feedback (RLHF).
- ▪The equivalence between DPO and RLHF is conditional, depending on the assumption that the RLHF-optimal policy must prefer human-preferred responses.
- ▪When this assumption fails, DPO may lead to undesirable outcomes, optimizing relative advantage instead of aligning with human preferences.
arXiv cs.AI files mainly under ai research. We currently carry 1,128 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | arXiv cs.AI |
| Canonical URL | https://arxiv.org/abs/2605.20834 |
| Publication time | Fri, 22 May 2026 00:00:00 -0400 |
| Retrieval time | 2026-05-22T04:02:00.009Z |
| Last seen | 2026-05-22T04:02:00.009Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | zjglxbD46GBQ |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.20834 (cs) [Submitted on 20 May 2026] Title:Conditional Equivalence of DPO and RLHF: Implicit Assumption, Failure Modes, and Provable Alignment Authors:Zhiqin Yang, Yonggang Zhang, Wei Xue, Dong Fang, Bo Han, Yike Guo View a PDF of the paper titled Conditional Equivalence of DPO and RLHF: Implicit Assumption, Failure Modes, and Provable Alignment, by Zhiqin Yang and 5 other authors View PDF HTML (experimental) Abstract:Direct Preference Optimization (DPO) has emerged as a popular alternative to Reinforcement Learning from Human Feedback (RLHF), offering theoretical equivalence with simpler implementation.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
