LIVE
Architectural tweaks may break conventional scaling law exponents16/09/26|OpenAI publishes a framework for reporting model misalignment16/09/26 · OpenAI|NVIDIA's Vera Rubin NVL72 Debuts in MLPerf Inference v6.116/09/26 · NVIDIA|OpenAI expands ChatGPT advertising with Sponsored Agents16/09/26 · OpenAI|OpenAI moves into advertising with 'Sponsored Agents'16/09/26 · OpenAI|Google DeepMind Introduces Gemini 3.8 Live and Its Extended Thinking Variant15/09/26 · Google DeepMind|What's at stake in AI's trillion-dollar infrastructure bet15/09/26|A Flaw in Chain-of-Thought Safety Monitoring14/09/26|Stellar Colosseum: A Multi-Agent System for Long-Horizon Mathematical Research14/09/26|Apple Code Hints Siri Could Be Swapped for ChatGPT or Claude14/09/26 · Apple|Anthropic says Houthi-linked actors used Claude Code for missile guidance software13/09/26 · Anthropic|Yoshua Bengio examines why AI agents lie, cheat and coordinate13/09/26|Architectural tweaks may break conventional scaling law exponents16/09/26|OpenAI publishes a framework for reporting model misalignment16/09/26 · OpenAI|NVIDIA's Vera Rubin NVL72 Debuts in MLPerf Inference v6.116/09/26 · NVIDIA|OpenAI expands ChatGPT advertising with Sponsored Agents16/09/26 · OpenAI|OpenAI moves into advertising with 'Sponsored Agents'16/09/26 · OpenAI|Google DeepMind Introduces Gemini 3.8 Live and Its Extended Thinking Variant15/09/26 · Google DeepMind|What's at stake in AI's trillion-dollar infrastructure bet15/09/26|A Flaw in Chain-of-Thought Safety Monitoring14/09/26|Stellar Colosseum: A Multi-Agent System for Long-Horizon Mathematical Research14/09/26|Apple Code Hints Siri Could Be Swapped for ChatGPT or Claude14/09/26 · Apple|Anthropic says Houthi-linked actors used Claude Code for missile guidance software13/09/26 · Anthropic|Yoshua Bengio examines why AI agents lie, cheat and coordinate13/09/26|
Research

ComPO: A Gradient-Free Approach to LLM Preference Alignment

Researchers introduce ComPO, a preference alignment method relying on comparison oracles instead of directly optimizing a differentiable loss, showing gains across several open-source model families.

September 16, 20263 min readPublished byarXiv

Direct preference alignment methods, such as DPO, are widely adopted to fine-tune large language models toward human preferences because of their computational efficiency. However, they suffer from a known issue called likelihood displacement: when the margin between preferred and rejected responses is small, optimization can inadvertently degrade overall model quality rather than improve it.

To address this, the authors propose ComPO (Comparison-based Preference Optimization), a zeroth-order method that avoids directly optimizing a differentiable loss over preference pairs. Instead, it extracts directional signals through comparison oracles, an approach related to gradient-free optimization techniques. The authors provide theoretical convergence guarantees for a basic offline scheme, under assumptions of smoothness, gradient sparsity, and compatibility between the oracle and a latent objective.

The paper also introduces an online variant, online ComPO, which keeps the offline comparison mechanism while leveraging unlabeled generations from the current policy to control reverse-KL divergence relative to a reference policy. A performance guarantee is established for a basic constrained scheme, conditional on local coverage and in-distribution pairwise reward accuracy — building on a coverage-based perspective of preference fine-tuning.

Experiments across Mistral, Llama, Gemma-2, Qwen3, and Gemma-3 models show improvements over existing direct alignment baselines, including length-controlled win rates. Pair-level diagnostics offer evidence consistent with mitigated likelihood displacement. The work represents a solid technical contribution within the ongoing effort to refine RLHF-style methods beyond DPO, though it is more incremental research than an immediate shift in industrial practice.

Tags
preference-alignmentrlhfzeroth-order-optimizationllm-trainingbenchmark

Read also