Asymmetric Capacity Allocation in Self-Refinement Pipelines

• URL: http://arxiv.org/abs/2608.21345v1
• Date: Fri, 21 Aug 2026 17:52:17 GMT
• Title: Asymmetric Capacity Allocation in Self-Refinement Pipelines
• Abstract summary: We present the first stage-wise model size study of the self-refinement pipeline on 5 benchmarks from different domains.<n>We conclude that larger generators and refiners generally improve the pipeline, whereas an undersized refiner can even harm performance.
• Score: 14.227188394452673
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Self-refinement, typically structured as generation, critique, and revision, is a widely adopted paradigm for improving LLM generation and serves as a core mechanism in many LLM agents. While the three stages involve different cognitive demands, most existing approaches conveniently treat the model size as an implementation detail rather than a subject of study, which may lead to a waste of resources. Little work has systematically examined how model size affects each stage or whether effective self-refinement requires equally capable models for generation, critique, and revision. We present the first stage-wise model size study of the self-refinement pipeline on 5 benchmarks from different domains using 6 model sizes of Qwen3 and 4 model sizes of Gemma 3. We conclude that larger generators and refiners generally improve the pipeline, whereas an undersized refiner can even harm performance. Second, performance is highly insensitive to the size of the critic, although including even a small critic consistently outperforms omitting critique altogether. Our findings demonstrate that model capacity should not be allocated uniformly across self-refinement pipelines. Instead, different stages exhibit distinct size scaling characteristics, providing practical guidance for designing more computationally efficient multi-stage language model systems.

Related papers

• Improving LLM-Generated Process Model Quality Through Reinforcement Learning: The Role of Reward Function Design [43.328395470827665]
  Large language models (LLMs) can generate BPMN process models from natural-language descriptions.<n>Supervised fine-tuning (SFT) limits their output quality to the patterns present in the training data.<n>We present a systematic investigation of reward function design for RL-based process model generation.
  arXiv  Detail & Related papers  (2026-07-07T11:53:02Z)
• Large Language Models as Discounted Bayesian Filters [14.164508061248775]
  We introduce a Bayesian filtering framework to evaluate online inference in Large Language Models (LLMs)<n>We find that while LLM belief updates resemble Bayesian posteriors, they are more accurately characterized by an exponential forgetting filter with a model-specific discount factor smaller than one.<n>Although inherent priors are often miscalibrated, the updating mechanism itself remains structured and principled.
  arXiv  Detail & Related papers  (2025-12-20T19:56:39Z)
• Which Layer Causes Distribution Deviation? Entropy-Guided Adaptive Pruning for Diffusion and Flow Models [77.55829017952728]
  EntPruner is an entropy-guided automatic progressive pruning framework for diffusion and flow models.<n>Experiments on DiT and SiT models demonstrate the effectiveness of EntPruner, achieving up to 2.22$times$ inference speedup.
  arXiv  Detail & Related papers  (2025-11-26T07:20:48Z)
• Can large language models assist choice modelling? Insights into prompting strategies and current models capabilities [0.0]
  Large Language Models (LLMs) are widely used to support various disciplines, yet their potential in choice modelling remains relatively unexplored.<n>This work examines the potential of LLMs as assistive agents in the specification and, where technically feasible, estimation of Multinomial Logit models.
  arXiv  Detail & Related papers  (2025-07-29T13:24:44Z)
• SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models [21.933379266533098]
  Large Language Models (LLMs) present a critical trade-off between inference quality and computational cost.<n>Existing serving strategies often employ fixed model scales or static two-stage speculative decoding.<n>This paper introduces systemname, a novel framework that reimagines LLM inference as an adaptive routing problem.
  arXiv  Detail & Related papers  (2025-05-12T15:46:28Z)
• Large EEG-U-Transformer for Time-Step Level Detection Without Pre-Training [1.3254304182988286]
  We propose a simple U-shaped model to efficiently learn representations by capturing both local and global features.<n>Compared to other window-level classification models, our method directly outputs predictions at the time-step level.<n>Our model won 1st place in the 2025 "seizure detection challenge" organized in the International Conference on Artificial Intelligence in Epilepsy and Other Neurological Disorders.
  arXiv  Detail & Related papers  (2025-04-01T01:33:42Z)
• Revisiting SMoE Language Models by Evaluating Inefficiencies with Task Specific Expert Pruning [78.72226641279863]
  Sparse Mixture of Expert (SMoE) models have emerged as a scalable alternative to dense models in language modeling. Our research explores task-specific model pruning to inform decisions about designing SMoE architectures. We introduce an adaptive task-aware pruning technique UNCURL to reduce the number of experts per MoE layer in an offline manner post-training.
  arXiv  Detail & Related papers  (2024-09-02T22:35:03Z)
• SMILE: Zero-Shot Sparse Mixture of Low-Rank Experts Construction From Pre-Trained Foundation Models [85.67096251281191]
  We present an innovative approach to model fusion called zero-shot Sparse MIxture of Low-rank Experts (SMILE) construction. SMILE allows for the upscaling of source models into an MoE model without extra data or further training. We conduct extensive experiments across diverse scenarios, such as image classification and text generation tasks, using full fine-tuning and LoRA fine-tuning.
  arXiv  Detail & Related papers  (2024-08-19T17:32:15Z)
• EMR-Merging: Tuning-Free High-Performance Model Merging [55.03509900949149]
  We show that Elect, Mask & Rescale-Merging (EMR-Merging) shows outstanding performance compared to existing merging methods. EMR-Merging is tuning-free, thus requiring no data availability or any additional training while showing impressive performance.
  arXiv  Detail & Related papers  (2024-05-23T05:25:45Z)
• The Role of Model Architecture and Scale in Predicting Molecular Properties: Insights from Fine-Tuning RoBERTa, BART, and LLaMA [0.0]
  This study introduces a systematic framework to compare the efficacy of Large Language Models (LLMs) for fine-tuning across various cheminformatics tasks. We assessed three well-known models-RoBERTa, BART, and LLaMA-on their ability to predict molecular properties. We found that LLaMA-based models generally offered the lowest validation loss, suggesting their superior adaptability across tasks and scales.
  arXiv  Detail & Related papers  (2024-05-02T02:20:12Z)
• Exploring the Limits of Domain-Adaptive Training for Detoxifying Large-Scale Language Models [84.30718841659531]
  We explore domain-adaptive training to reduce the toxicity of language models. For the training corpus, we propose to leverage the generative power of LMs. We then comprehensively study LMs with parameter sizes ranging from 126M up to 530B, a scale that has never been studied before.
  arXiv  Detail & Related papers  (2022-02-08T22:10:40Z)
• Reinforcement Learning as One Big Sequence Modeling Problem [84.84564880157149]
  Reinforcement learning (RL) is typically concerned with estimating single-step policies or single-step models. We view RL as a sequence modeling problem, with the goal being to predict a sequence of actions that leads to a sequence of high rewards.
  arXiv  Detail & Related papers  (2021-06-03T17:58:51Z)
• How Faithful is your Synthetic Data? Sample-level Metrics for Evaluating and Auditing Generative Models [95.8037674226622]
  We introduce a 3-dimensional evaluation metric that characterizes the fidelity, diversity and generalization performance of any generative model in a domain-agnostic fashion. Our metric unifies statistical divergence measures with precision-recall analysis, enabling sample- and distribution-level diagnoses of model fidelity and diversity.
  arXiv  Detail & Related papers  (2021-02-17T18:25:30Z)

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.