From "Weak" Signals to Strong Models: Preference Delta Aggregation with LoRA Merging

• URL: http://arxiv.org/abs/2606.00357v1
• Date: Fri, 29 May 2026 21:00:29 GMT
• Title: From "Weak" Signals to Strong Models: Preference Delta Aggregation with LoRA Merging
• Abstract summary: Training strong large language models (LLMs) requires high-quality supervision, which is often scarce.<n>Recent work shows that paired preference data from weak-weaker model pairs can provide an effective supervision signal through relative quality deltas.<n>This motivates a key research question: can multiple "weak" signals be constructively aggregated for improving strong models (e.g., Qwen3 8B)?<n>PDA is the first framework that derives a preference delta from each weak-weaker model pair, instantiates it as a LoRA adapter learned through preference optimization, and aggregates the resulting deltas via LoRA merging.
• Score: 26.22748334252526
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Training strong large language models (LLMs) requires high-quality supervision, which is often scarce. Recent work shows that paired preference data from weak-weaker model pairs (e.g., Qwen3 4B over 1.7B), despite the limited quality of individual responses, can provide an effective supervision signal through relative quality deltas, which we term a "weak" signal. This motivates a key research question: can multiple "weak" signals be constructively aggregated for improving strong models (e.g., Qwen3 8B)? To this end, we propose Preference Delta Aggregation (PDA), the first framework that derives a preference delta from each weak-weaker model pair, instantiates it as a LoRA adapter learned through preference optimization, and aggregates the resulting deltas via LoRA merging. To further mitigate directional interference during LoRA merging, we introduce Geometric Alignment Merging (GAM), a geometry-aware merging method that aligns adapter subspaces before aggregation, enabling more robust composition of diverse deltas. Evaluations on knowledge reasoning and agentic search benchmarks show that aggregating multiple "weak" signals pushes performance beyond any single signal, with further gains as additional signals are incorporated. Correspondingly, PDA with GAM improves the strong model by 6.8 and 7.3 points on average for knowledge reasoning and agentic search, respectively. It outperforms all single-delta and multi-delta baselines, exceeding the best single-delta baseline by 2.1 and 4.3 points. Further analysis attributes these gains to the effective composition of complementary capabilities encoded across distinct preference deltas.

Related papers

• AlignDiff: Exploiting Model-Intrinsic Information for Better Preference Data Selection [53.61691036434302]
  We propose AlignDiff, a preference data filtering framework driven by intrinsic model signals.<n>AlpacaEval 2.0, Arena-Hard, and MT-Bench benchmarks are evaluated.<n>We conduct comprehensive ablation studies to validate the effectiveness of AlignDiff.
  arXiv  Detail & Related papers  (2026-09-05T05:51:23Z)
• SignalReasoner: Assessing the Upper Bound of 3B Models for Signal Mathematical Reasoning [0.0]
  Post-training with supervised chain-of-thought fine-tuning has substantially improved the mathematical reasoning capabilities of large language models (LLMs)<n>This report investigates reinforcement fine-tuning strategies for adapting Qwen2.5-3B-Base to graduate-level signal mathematical problems from WirelessMATHBench-XL.
  arXiv  Detail & Related papers  (2026-08-18T02:52:41Z)
• Statistically Grounded Sparse-Feature Interventions for Activation-Space Control in Large Language Models [2.464003792743989]
  Activation steering offers a lightweight alternative to fine-tuning for behavioral control of large language models.<n>We introduce a transparent SAE-feature steering pipeline that ranks sparse features through an unweighted Borda consensus.<n>Across three Gemma-family models, four behavioral domains, and 356 layer-strength configurations, the method produces measurable domain-specific shifts.
  arXiv  Detail & Related papers  (2026-06-05T19:27:31Z)
• TaDA: Calibrated Probe Gating for Task-Domain LoRA Merging [6.356259006504384]
  We argue that task and domain adapters exhibit a consistent depth-dependent asymmetry across transformer architectures.<n>We propose a training-free algorithm that exploits this structure through calibrated probe-guided per-layer gating and per-component subspace-aware merging.<n>$textbfTaDA$ produces a standard rank-$r$ LoRA adapter with zero inference overhead.
  arXiv  Detail & Related papers  (2026-06-03T15:39:37Z)
• When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors [66.18091962164219]
  Existing metrics fail to distinguish mandatory behaviors required for task success from non-mandatory patterns that reflect a model's autonomous preferences.<n>We propose two complementary metrics to isolate non-mandatory behavioral patterns: textbfResponse Pattern Similarity (RPS) for verbal alignment and textbfAction Graph Similarity (AGS) for tool-use habits modeled as directed graphs.
  arXiv  Detail & Related papers  (2026-04-23T03:48:56Z)
• PaTaRM: Bridging Pairwise and Pointwise Signals via Preference-Aware Task-Adaptive Reward Modeling [19.258007121955924]
  Preference-Aware Task-Aware Reward Model (PaTaRM) is a unified framework that integrates a preference-aware reward mechanism with dynamic rubric adaptation.<n>PaTaRM boosts downstream RLHF performance, with an average improvement of 13.6% across IFEval and InFoBench benchmarks.
  arXiv  Detail & Related papers  (2025-10-28T09:43:47Z)
• Merge and Guide: Unifying Model Merging and Guided Decoding for Controllable Multi-Objective Generation [49.98025799046136]
  We introduce Merge-And-GuidE, a two-stage framework that leverages model merging for guided decoding.<n>In Stage 1, MAGE resolves a compatibility problem between the guidance and base models.<n>In Stage 2, we merge explicit and implicit value models into a unified guidance proxy, which then steers the decoding of the base model from Stage 1.
  arXiv  Detail & Related papers  (2025-10-04T11:10:07Z)
• Optimal Control Meets Flow Matching: A Principled Route to Multi-Subject Fidelity [35.95129874095729]
  Text-to-image (T2I) models excel on single-entity prompts but struggle with multi-subject descriptions.<n>We introduce the first theoretical framework with principled optimizable objective for steering sampling dynamics toward multi-subject fidelity.
  arXiv  Detail & Related papers  (2025-10-02T17:59:58Z)
• Efficient Modular Learning through Naive LoRA Summation: Leveraging Orthogonality in High-Dimensional Models [1.0923877073891446]
  Low-Rank Adaptation (LoRA) stores parameter deltas as the product of two small matrices.<n>Naive summation requires no additional training, can be applied in seconds, and achieves performance comparable to models trained on merged data.
  arXiv  Detail & Related papers  (2025-08-16T08:49:02Z)
• The Delta Learning Hypothesis: Preference Tuning on Weak Data can Yield Strong Gains [50.66245575710432]
  We show that paired preference data consisting of individually weak data points can enable gains beyond the strength of each individual data point.<n>Our work shows that models can learn surprisingly well from paired data that might typically be considered weak.
  arXiv  Detail & Related papers  (2025-07-08T17:14:44Z)
• Anchored Preference Optimization and Contrastive Revisions: Addressing Underspecification in Alignment [57.03947082589616]
  Large Language Models (LLMs) are often aligned using contrastive alignment objectives and preference pair datasets. We study this and find that preference data gives a better learning signal when the underlying responses are contrastive. We introduce Contrastive Learning from AI Revisions (CLAIR), a data-creation method which leads to more contrastive preference pairs. Our best model, trained on 32K CLAIR preferences with APO, improves Llama-3-8B-Instruct by 7.65%, closing the gap with GPT4-turbo by 45%.
  arXiv  Detail & Related papers  (2024-08-12T16:24:51Z)
• DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling [24.270321913746233]
  We propose a new model merging technique, Drop and rEscaLe via sampLing with mAgnitude (DELLA-Merging), that employs a novel pruning technique, MAGPRUNE. MAGPRUNE first ranks the parameters in order of their magnitude and assigns higher dropout probabilities (p) to parameters with lower ranks corresponding to lower magnitudes.
  arXiv  Detail & Related papers  (2024-06-17T15:02:45Z)
• AffineGlue: Joint Matching and Robust Estimation [74.04609046690913]
  We propose AffineGlue, a method for joint two-view feature matching and robust estimation. AffineGlue selects potential matches from one-to-many correspondences to estimate minimal models. Guided matching is then used to find matches consistent with the model, suffering less from the ambiguities of one-to-one matches.
  arXiv  Detail & Related papers  (2023-07-28T08:05:36Z)
• AvgOut: A Simple Output-Probability Measure to Eliminate Dull Responses [97.50616524350123]
  We build dialogue models that are dynamically aware of what utterances or tokens are dull without any feature-engineering. The first model, MinAvgOut, directly maximizes the diversity score through the output distributions of each batch. The second model, Label Fine-Tuning (LFT), prepends to the source sequence a label continuously scaled by the diversity score to control the diversity level. The third model, RL, adopts Reinforcement Learning and treats the diversity score as a reward signal.
  arXiv  Detail & Related papers  (2020-01-15T18:32:06Z)

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.