ELSPR: Evaluator LLM Training Data Self-Purification on Non-Transitive Preferences via Tournament Graph Reconstruction

• URL: http://arxiv.org/abs/2505.17691v1
• Date: Fri, 23 May 2025 10:00:03 GMT
• Title: ELSPR: Evaluator LLM Training Data Self-Purification on Non-Transitive Preferences via Tournament Graph Reconstruction
• Authors: Yan Yu, Yilun Liu, Minggui He, Shimin Tao, Weibin Meng, Xinhua Yang, Li Zhang, Hongxia Ma, Chang Su, Hao Yang, Fuliang Li,
• Abstract summary: Low-quality training data may reduce the transitivity of preferences generated by the Evaluator LLM.<n>We quantify non-transitivity and introduce directed graph structural entropy to measure the overall clarity of preferences.<n>Experiments demonstrate that models fine-tuned with filtered data reduce non-transitivity by 13.78%, decrease structural entropy by 0.0879, and align more closely with human evaluators.
• Score: 12.834553446564634
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) are widely used as evaluators for open-ended tasks, while previous research has emphasized biases in LLM evaluations, the issue of non-transitivity in pairwise comparisons remains unresolved: non-transitive preferences for pairwise comparisons, where evaluators prefer A over B, B over C, but C over A. Our results suggest that low-quality training data may reduce the transitivity of preferences generated by the Evaluator LLM. To address this, We propose a graph-theoretic framework to analyze and mitigate this problem by modeling pairwise preferences as tournament graphs. We quantify non-transitivity and introduce directed graph structural entropy to measure the overall clarity of preferences. Our analysis reveals significant non-transitivity in advanced Evaluator LLMs (with Qwen2.5-Max exhibiting 67.96%), as well as high entropy values (0.8095 for Qwen2.5-Max), reflecting low overall clarity of preferences. To address this issue, we designed a filtering strategy, ELSPR, to eliminate preference data that induces non-transitivity, retaining only consistent and transitive preference data for model fine-tuning. Experiments demonstrate that models fine-tuned with filtered data reduce non-transitivity by 13.78% (from 64.28% to 50.50%), decrease structural entropy by 0.0879 (from 0.8113 to 0.7234), and align more closely with human evaluators (human agreement rate improves by 0.6% and Spearman correlation increases by 0.01).

Related papers

• Advancing Tabular Stroke Modelling Through a Novel Hybrid Architecture and Feature-Selection Synergy [0.9999629695552196]
  The present work develops and validates a data-driven and interpretable machine-learning framework designed to predict strokes.<n>Ten routinely gathered demographic, lifestyle, and clinical variables were sourced from a public cohort of 4,981 records.<n>The proposed model achieved an accuracy rate of 97.2% and an F1-score of 97.15%, indicating a significant enhancement compared to the leading individual model.
  arXiv  Detail & Related papers  (2025-05-18T21:46:45Z)
• Leveraging Robust Optimization for LLM Alignment under Distribution Shifts [52.983390470606146]
  Preference alignment methods are increasingly critical for steering large language models to generate outputs consistent with human values.<n>We propose a novel distribution-aware optimization framework that improves preference alignment despite such shifts.
  arXiv  Detail & Related papers  (2025-04-08T09:14:38Z)
• Gradient Extrapolation for Debiased Representation Learning [7.183424522250937]
  Gradient Extrapolation for Debiased Representation Learning (GERNE) is designed to learn debiased representations in both known and unknown attribute training cases.<n>GERNE can serve as a general framework for debiasing with methods, such as ERM, reweighting, and resampling, being shown as special cases.<n>The proposed approach is validated on five vision and one NLP benchmarks, demonstrating competitive and often superior performance compared to state-of-the-art baseline methods.
  arXiv  Detail & Related papers  (2025-03-17T14:48:57Z)
• Dynamic Noise Preference Optimization for LLM Self-Improvement via Synthetic Data [51.62162460809116]
  We introduce Dynamic Noise Preference Optimization (DNPO) to ensure consistent improvements across iterations.<n>In experiments with Zephyr-7B, DNPO consistently outperforms existing methods, showing an average performance boost of 2.6%.<n> DNPO shows a significant improvement in model-generated data quality, with a 29.4% win-loss rate gap compared to the baseline in GPT-4 evaluations.
  arXiv  Detail & Related papers  (2025-02-08T01:20:09Z)
• EACO: Enhancing Alignment in Multimodal LLMs via Critical Observation [58.546205554954454]
  We propose Enhancing Alignment in MLLMs via Critical Observation (EACO)<n>EACO aligns MLLMs by self-generated preference data using only 5k images economically.<n>EACO reduces the overall hallucinations by 65.6% on HallusionBench and improves the reasoning ability by 21.8% on MME-Cognition.
  arXiv  Detail & Related papers  (2024-12-06T09:59:47Z)
• Geometric-Averaged Preference Optimization for Soft Preference Labels [78.2746007085333]
  Many algorithms for aligning LLMs with human preferences assume that human preferences are binary and deterministic.<n>In this work, we introduce the distributional soft preference labels and improve Direct Preference Optimization (DPO) with a weighted geometric average of the LLM output likelihood in the loss function.
  arXiv  Detail & Related papers  (2024-09-10T17:54:28Z)
• Crafting Efficient Fine-Tuning Strategies for Large Language Models [2.633490094119608]
  Fine-tuning large language models (LLMs) with as few as 200 samples can improve model accuracy from 70% to 88% in a product attribute extraction task. A bayesian hyperparameter optimization method, which evaluates models at 20% of total training time, correlates strongly with final model performance. This approach led to a 2% improvement in accuracy over baseline models when evaluated on an independent test set.
  arXiv  Detail & Related papers  (2024-07-18T21:36:00Z)
• Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs [54.05511925104712]
  We propose a simple, effective, and data-efficient method called Step-DPO. Step-DPO treats individual reasoning steps as units for preference optimization rather than evaluating answers holistically. Our findings demonstrate that as few as 10K preference data pairs and fewer than 500 Step-DPO training steps can yield a nearly 3% gain in accuracy on MATH for models with over 70B parameters.
  arXiv  Detail & Related papers  (2024-06-26T17:43:06Z)
• Spread Preference Annotation: Direct Preference Judgment for Efficient LLM Alignment [72.99676237703099]
  We propose a new framework that boosts the alignment of large language models with human preferences.<n>Our key idea is leveraging the human prior knowledge within the small (seed) data.<n>We introduce a noise-aware preference learning algorithm to mitigate the risk of low quality within generated preference data.
  arXiv  Detail & Related papers  (2024-06-06T18:01:02Z)
• Dissecting Human and LLM Preferences [80.55271307662365]
  We find that humans are less sensitive to errors, favor responses that support their stances, and show clear dislike when models admit their limits. advanced LLMs like GPT-4-Turbo emphasize correctness, clarity, and harmlessness more. We show that preference-based evaluation can be intentionally manipulated.
  arXiv  Detail & Related papers  (2024-02-17T14:34:31Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.