Human Feedback Types
strongPairwise Preference
Directly usable for protocol triage.
"Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences."
HFEPX · Eval paper review
Less is More: Improving LLM Alignment via Preference Data Selection
Xun Deng, Han Zhong, Rui Ai, Fuli Feng +2 more
Published
Feb 20, 2025
Citations
0
Trust level
Moderate
Usefulness score
50/100 (Medium)
Extraction confidence
55% (Moderate)
Derived from extracted protocol signals and abstract evidence.
Rater population
Not reported
Signals refreshed
Feb 15, 2026
Should you rely on this paper?
This paper has useful evaluation signal, but protocol completeness is partial; pair it with related papers before deciding implementation strategy.
Use this for comparison and orientation, not as your only source.
Low-signal caution for protocol decisions
Use this page for context, then validate protocol choices against stronger HFEPX references before implementation decisions.
- The abstract does not clearly describe the evaluation setup.
Best use
Secondary protocol comparison source
Use if you need
Background context only.
What to verify
Validate the evaluation procedure and quality controls in the full paper before operational use.
Main weakness
The abstract does not clearly describe the evaluation setup.
Human feedback signal
Detected
From extracted signals
Evaluation signal
Weak or implicit
Validate from full paper
Usefulness for eval research
50/100
Moderate-confidence candidate
Useful as a secondary reference; validate protocol details against neighboring papers.
If you are doing eval pipeline work, start here
Abstract
Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences. While prior work mainly extends DPO from the aspect of the objective function, we instead improve DPO from the largely overlooked but critical aspect of data selection. Specifically, we address the issue of parameter shrinkage caused by noisy data by proposing a novel margin-maximization principle for dataset curation in DPO training. To further mitigate the noise in different reward models, we propose a Bayesian Aggregation approach that unifies multiple margin sources (external and implicit) into a single preference probability. Extensive experiments in diverse settings demonstrate the consistently high data efficiency of our approach. Remarkably, by using just 10\% of the Ultrafeedback dataset, our approach achieves 3\% to 8\% improvements across various Llama, Mistral, and Qwen models on the AlpacaEval2 benchmark. Furthermore, our approach seamlessly extends to iterative DPO, yielding a roughly 3\% improvement with 25\% online data, revealing the high redundancy in this presumed high-quality data construction manner. These results highlight the potential of data selection strategies for advancing preference optimization.
What we could verify
These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.
Evaluation Modes
missingNone explicit
Validate eval design from full paper text.
"Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences."
Quality Controls
missingNot reported
No explicit QC controls found.
"Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences."
Benchmarks / Datasets
strongAlpacaEval 2.0
Useful for quick benchmark comparison.
"Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences."
Reported Metrics
missingNot extracted
No metric anchors detected.
"Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences."
Benchmarks and datasets
AlpacaEval 2.0
Reported metrics
No metric terms were extracted from the available abstract.
Human feedback details
- Uses human feedback
- Yes
- Feedback types
- Pairwise Preference
- Rater population
- Not reported
- Expertise required
- General
Evaluation details
- Evaluation modes
- None
- Agentic eval
- None
- Quality controls
- Not reported
- Evidence quality
- Moderate
- Use this page as
- Secondary protocol comparison source
Research brief
Metadata summary
Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences.
Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.
Key takeaways
- Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences.
- While prior work mainly extends DPO from the aspect of the objective function, we instead improve DPO from the largely overlooked but critical aspect of data selection.
- Specifically, we address the issue of parameter shrinkage caused by noisy data by proposing a novel margin-maximization principle for dataset curation in DPO training.
Researcher actions
- Compare this paper against nearby papers in the same arXiv category before using it for protocol decisions.
- Check the full text for explicit evaluation design choices (raters, protocol, and metrics).
- Use related-paper links to find stronger protocol-specific references.
Caveats
- Generated from abstract + metadata only; no PDF parsing.
- Signals below are heuristic and may miss details reported outside the abstract.
Recommended queries
Contribution summary
- Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences.
- To further mitigate the noise in different reward models, we propose a Bayesian Aggregation approach that unifies multiple margin sources (external and implicit) into a single preference probability.
- Remarkably, by using just 10\% of the Ultrafeedback dataset, our approach achieves 3\% to 8\% improvements across various Llama, Mistral, and Qwen models on the AlpacaEval2 benchmark.
Why it matters for eval
- Direct Preference Optimization (DPO) has emerged as a promising approach for aligning large language models with human preferences.
- To further mitigate the noise in different reward models, we propose a Bayesian Aggregation approach that unifies multiple margin sources (external and implicit) into a single preference probability.
Researcher checklist
-
Human feedback protocol is explicit
Detected: Pairwise Preference
-
Evaluation mode is explicit
No clear evaluation mode extracted.
-
Quality control reporting appears
No calibration/adjudication/IAA control explicitly detected.
-
Benchmark or dataset anchors are present
Detected: AlpacaEval 2.0
-
Metric reporting is present
No metric terms extracted.