Hierarchical Alignment: Surgical Fine-Tuning via Functional Layer Specialization in Large Language Models

• URL: http://arxiv.org/abs/2510.12044v1
• Date: Tue, 14 Oct 2025 00:58:34 GMT
• Title: Hierarchical Alignment: Surgical Fine-Tuning via Functional Layer Specialization in Large Language Models
• Authors: Yukun Zhang, Qi Dong,
• Abstract summary: We introduce Hierarchical Alignment, a novel method that applies targeted DPO to distinct functional blocks of a model's layers.<n>Specifically, aligning the local layers (Local-Align) enhances grammatical fluency.<n> aligning the global layers (Global-Align) improves factual consistency as hypothesized but also proves to be the most effective strategy for enhancing logical coherence.
• Score: 4.935224714809964
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Existing alignment techniques for Large Language Models (LLMs), such as Direct Preference Optimization (DPO), typically treat the model as a monolithic entity, applying uniform optimization pressure across all layers. This approach overlooks the functional specialization within the Transformer architecture, where different layers are known to handle distinct tasks from syntax to abstract reasoning. In this paper, we challenge this one-size-fits-all paradigm by introducing Hierarchical Alignment, a novel method that applies targeted DPO to distinct functional blocks of a model's layers: local (syntax), intermediate (logic), and global (factuality). Through a series of controlled experiments on state-of-the-art models like Llama-3.1-8B and Qwen1.5-7B using LoRA for surgical fine-tuning, our results, evaluated by a powerful LLM-as-Judge, demonstrate significant and predictable improvements. Specifically, aligning the local layers (Local-Align) enhances grammatical fluency. More importantly, aligning the global layers (Global-Align) not only improves factual consistency as hypothesized but also proves to be the most effective strategy for enhancing logical coherence, outperforming all baselines. Critically, all hierarchical strategies successfully avoid the "alignment tax" observed in standard DPO, where gains in fluency come at the cost of degraded logical reasoning. These findings establish a more resource-efficient, controllable, and interpretable path for model alignment, highlighting the immense potential of shifting from monolithic optimization to structure-aware surgical fine-tuning to build more advanced and reliable LLMs.

Related papers

• Multi-Objective Hierarchical Optimization with Large Language Models [41.41567058185742]
  Large Language Models (LLMs) are not the off-the-shelf choice to drive multi-objective optimization yet.<n>In this paper, we close this gap by leveraging LLMs as surrogate models and candidate samplers inside a structured hierarchical search strategy.
  arXiv  Detail & Related papers  (2026-01-20T12:10:13Z)
• Closed-Loop LLM Discovery of Non-Standard Channel Priors in Vision Models [48.83701310501069]
  Large Language Models (LLMs) offer a transformative approach to Neural Architecture Search (NAS)<n>We formulate the search as a sequence of conditional code generation tasks, where an LLM refines architectural specifications based on performance telemetry.<n>We generate a vast corpus of valid, shape-consistent architectures via Abstract Syntax Tree (AST) mutations.<n> Experimental results on CIFAR-100 validate the efficacy of this approach, demonstrating that the model yields statistically significant improvements in accuracy.
  arXiv  Detail & Related papers  (2026-01-13T13:00:30Z)
• Hierarchy-Aware Fine-Tuning of Vision-Language Models [18.244518940229202]
  Vision-Language Models learn powerful multimodal representations through large-scale image-text pretraining.<n>Standard approaches treat labels as flat categories and require full fine-tuning, which is expensive and produces inconsistent predictions.<n>We propose an efficient hierarchy-aware fine-tuning framework that updates a few parameters while enforcing structural consistency.
  arXiv  Detail & Related papers  (2025-12-25T06:44:33Z)
• Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies [30.35690865689118]
  Existing reinforcement learning approaches treat large language models (LLMs) as a single unified policy, overlooking their internal mechanisms.<n>We decompose the language model policy by leveraging the intrinsic split of the Transformer residual stream and the equivalence between the composition of hidden states with the unembedding matrix.<n>This decomposition reveals Internal Layer Policies, corresponding to contributions from individual layers, and Internal Modular Policies, which align with the self-attention and feed-forward network.
  arXiv  Detail & Related papers  (2025-12-22T18:51:48Z)
• Multiscale Aggregated Hierarchical Attention (MAHA): A Game Theoretic and Optimization Driven Approach to Efficient Contextual Modeling in Large Language Models [0.0]
  Multiscale Aggregated Hierarchical Attention (MAHA) is a novel architectural framework that reformulates the attention mechanism through hierarchical decomposition and mathematically rigorous aggregation.<n>MAHA dynamically partitions the input sequence into hierarchical scales via learnable downsampling operators.<n> Experimental evaluations demonstrate that MAHA achieves superior scalability; empirical FLOPs analysis confirms an 81% reduction in computational cost at a sequence length of 4096 compared to standard attention.
  arXiv  Detail & Related papers  (2025-12-16T21:27:21Z)
• Hierarchical LoRA MoE for Efficient CTR Model Scaling [56.608809143548946]
  HiLoMoE is a hierarchical LoRA MoE framework that enables holistic scaling in a parameter-efficient manner.<n>Unlike conventional stacking, HiLoMoE routes based on prior layer scores rather than outputs, allowing all layers to execute in parallel.
  arXiv  Detail & Related papers  (2025-10-12T03:54:11Z)
• TreeLoRA: Efficient Continual Learning via Layer-Wise LoRAs Guided by a Hierarchical Gradient-Similarity Tree [52.44403214958304]
  In this paper, we introduce TreeLoRA, a novel approach that constructs layer-wise adapters by leveraging hierarchical gradient similarity.<n>To reduce the computational burden of task similarity estimation, we employ bandit techniques to develop an algorithm based on lower confidence bounds.<n> experiments on both vision transformers (ViTs) and large language models (LLMs) demonstrate the effectiveness and efficiency of our approach.
  arXiv  Detail & Related papers  (2025-06-12T05:25:35Z)
• IMPROVE: Iterative Model Pipeline Refinement and Optimization Leveraging LLM Experts [28.9807389592324]
  Large language model (LLM) agents have emerged as a promising solution to automate the workflow of machine learning.<n>We introduce Iterative Refinement, a novel strategy for LLM-driven ML pipeline design inspired by how human ML experts iteratively refine models.<n>By systematically updating individual components based on real training feedback, Iterative Refinement improves overall model performance.
  arXiv  Detail & Related papers  (2025-02-25T01:52:37Z)
• Understanding Layer Significance in LLM Alignment [23.582520695083588]
  We propose identifying which layers within large language models are most critical to the alignment process.<n> Experimental results reveal that, despite substantial differences in alignment datasets, the important layers of a model exhibit nearly 90% overlap.<n>The results also indicate that freezing non-essential layers improves overall model performance, while selectively tuning the most critical layers significantly enhances fine-tuning efficiency with minimal performance loss.
  arXiv  Detail & Related papers  (2024-10-23T13:47:05Z)
• How to Make LLMs Strong Node Classifiers? [70.14063765424012]
  Language Models (LMs) are challenging the dominance of domain-specific models, such as Graph Neural Networks (GNNs) and Graph Transformers (GTs)<n>We propose a novel approach that empowers off-the-shelf LMs to achieve performance comparable to state-of-the-art (SOTA) GNNs on node classification tasks.
  arXiv  Detail & Related papers  (2024-10-03T08:27:54Z)
• Entropy-Regularized Token-Level Policy Optimization for Language Agent Reinforcement [67.1393112206885]
  Large Language Models (LLMs) have shown promise as intelligent agents in interactive decision-making tasks. We introduce Entropy-Regularized Token-level Policy Optimization (ETPO), an entropy-augmented RL method tailored for optimizing LLMs at the token level. We assess the effectiveness of ETPO within a simulated environment that models data science code generation as a series of multi-step interactive tasks.
  arXiv  Detail & Related papers  (2024-02-09T07:45:26Z)
• Language Models as Hierarchy Encoders [22.03504018330068]
  We introduce a novel approach to re-train transformer encoder-based LMs as Hierarchy Transformer encoders (HiTs) Our method situates the output embedding space of pre-trained LMs within a Poincar'e ball with a curvature that adapts to the embedding dimension. We evaluate HiTs against pre-trained LMs, standard fine-tuned LMs, and several hyperbolic embedding baselines.
  arXiv  Detail & Related papers  (2024-01-21T02:29:12Z)
• ECoFLaP: Efficient Coarse-to-Fine Layer-Wise Pruning for Vision-Language Models [70.45441031021291]
  Large Vision-Language Models (LVLMs) can understand the world comprehensively by integrating rich information from different modalities. LVLMs are often problematic due to their massive computational/energy costs and carbon consumption. We propose Efficient Coarse-to-Fine LayerWise Pruning (ECoFLaP), a two-stage coarse-to-fine weight pruning approach for LVLMs.
  arXiv  Detail & Related papers  (2023-10-04T17:34:00Z)
• Optimization-Inspired Learning with Architecture Augmentations and Control Mechanisms for Low-Level Vision [74.9260745577362]
  This paper proposes a unified optimization-inspired learning framework to aggregate Generative, Discriminative, and Corrective (GDC) principles. We construct three propagative modules to effectively solve the optimization models with flexible combinations. Experiments across varied low-level vision tasks validate the efficacy and adaptability of GDC.
  arXiv  Detail & Related papers  (2020-12-10T03:24:53Z)

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.