RePo: Language Models with Context Re-Positioning

• URL: http://arxiv.org/abs/2512.14391v1
• Date: Tue, 16 Dec 2025 13:30:30 GMT
• Title: RePo: Language Models with Context Re-Positioning
• Abstract summary: In-context learning is fundamental to modern Large Language Models (LLMs)<n> prevailing architectures impose a rigid and fixed contextual structure by assigning linear or constant positional indices.<n>We propose RePo, a novel mechanism that reduces extraneous load via context re-positioning.
• Score: 10.269249887819988
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In-context learning is fundamental to modern Large Language Models (LLMs); however, prevailing architectures impose a rigid and fixed contextual structure by assigning linear or constant positional indices. Drawing on Cognitive Load Theory (CLT), we argue that this uninformative structure increases extraneous cognitive load, consuming finite working memory capacity that should be allocated to deep reasoning and attention allocation. To address this, we propose RePo, a novel mechanism that reduces extraneous load via context re-positioning. Unlike standard approaches, RePo utilizes a differentiable module, $f_φ$, to assign token positions that capture contextual dependencies, rather than replying on pre-defined integer range. By continually pre-training on the OLMo-2 1B backbone, we demonstrate that RePo significantly enhances performance on tasks involving noisy contexts, structured data, and longer context length, while maintaining competitive performance on general short-context tasks. Detailed analysis reveals that RePo successfully allocate higher attention to distant but relevant information, assign positions in dense and non-linear space, and capture the intrinsic structure of the input context. Our code is available at https://github.com/SakanaAI/repo.

Related papers

• Information Abundance Paradox: Long-Context Training Undermines Parametric Knowledge [72.54679781634037]
  We study how the context window shapes a model's mode of learning, shifting it between parametric internalization and contextualization.<n>We propose the Information Paradox, which hypothesizes that abundant relevant information in the training context can reduce the incentive to encode that information parametrically.
  arXiv  Detail & Related papers  (2026-08-12T16:13:05Z)
• RepoReasoner: Evaluating Repository-Level Code Reasoning Ability of Long-Context Language Models [54.53236295237077]
  We introduce RepoReasoner, a benchmark for evaluating repository-level code reasoning.<n> Output Prediction measures fine-grained, stateful execution reasoning across files.<n>Call Chain Prediction evaluates high-level architectural dependency understanding under noisy context.
  arXiv  Detail & Related papers  (2026-07-28T17:12:41Z)
• MIITA: Memory-Induced Inference-Time Adaptation for Continual Learning with Small Language Models [64.2820121997882]
  MIITA is a Memory-Induced Inference-Time Adaptation framework for supervised CL under constrained storage.<n>It stores supervised experiences as compact correction-direction prototypes with semantic anchors, and retrieves them at inference time using semantic and uncertainty-based cues.<n>Experiments show that MIITA consistently improves final performance and mitigates forgetting under fixed memory budgets.
  arXiv  Detail & Related papers  (2026-05-20T03:03:04Z)
• GRAVITY: Architecture-Agnostic Structured Anchoring for Long-Horizon Conversational Memory [38.10762218232756]
  We introduce GRAVITY, a plug-and-play structured memory module.<n>It extracts three knowledge representations from raw conversational utterances.<n>It injects these representations into the host system's prompt as structured anchoring contexts.
  arXiv  Detail & Related papers  (2026-05-03T02:55:46Z)
• Knowledge Capsules: Structured Nonparametric Memory Units for LLMs [0.6524460254566904]
  Large language models (LLMs) encode knowledge in parametric weights, making it costly to update or extend without retraining.<n>Retrieval-augmented generation (RAG) mitigates this limitation by appending retrieved text to the input, but operates purely through context expansion.<n>We propose Knowledge Capsules, structured nonparametric memory units that represent normalized relational knowledge.
  arXiv  Detail & Related papers  (2026-04-22T12:21:00Z)
• From Context to EDUs: Faithful and Structured Context Compression via Elementary Discourse Unit Decomposition [46.36937947958481]
  We introduce a novel explicit compression framework designed to preserve both global structure and fine-grained details.<n>Our approach reformulates a structural context compression as a structure-then-select process.<n>Our method achieves state-of-the-art structural prediction accuracy and significantly outperforms frontier LLMs.
  arXiv  Detail & Related papers  (2025-12-16T09:52:58Z)
• Beyond Real: Imaginary Extension of Rotary Position Embeddings for Long-Context LLMs [72.8830548005884]
  Rotary Position Embeddings (RoPE) have become a standard for encoding sequence order in Large Language Models.<n>Standard implementations utilize only the real component of the complex-valued dot product for attention score calculation.<n>We propose an extension that re-incorporates this imaginary component.
  arXiv  Detail & Related papers  (2025-12-08T12:59:54Z)
• Concept than Document: Context Compression via AMR-based Conceptual Entropy [21.954536296551]
  Large Language Models (LLMs) face information overload when handling long contexts, particularly in Retrieval-Augmented Generation (RAG)<n>We propose an unsupervised context compression framework that exploits Abstract Representation (AMR) graphs to preserve semantically essential information while filtering out irrelevant text.
  arXiv  Detail & Related papers  (2025-11-24T07:08:02Z)
• Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models [18.829572148850563]
  We introduce ACE (Agentic Context Engineering), a framework that treats contexts as evolving playbooks.<n>Across agent and domain-specific benchmarks, ACE consistently outperforms strong baselines.<n> ACE could adapt effectively without labeled supervision and instead by leveraging natural execution feedback.
  arXiv  Detail & Related papers  (2025-10-06T09:30:18Z)
• CoT Referring: Improving Referring Expression Tasks with Grounded Reasoning [67.18702329644526]
  CoT Referring enhances model reasoning across modalities through a structured, chain-of-thought training data structure.<n>We restructure the training data to enforce a new output form, providing new annotations for existing datasets.<n>We also integrate detection and segmentation capabilities into a unified MLLM framework, training it with a novel adaptive weighted loss to optimize performance.
  arXiv  Detail & Related papers  (2025-10-03T08:50:21Z)
• Scalable In-Context Q-Learning [68.9917436397079]
  We propose textbfScalable textbfIn-textbfContext textbfQ-textbfLearning (textbfSICQL) to steer in-context reinforcement learning.<n>textbfSICQL harnesses dynamic programming and world modeling to steer ICRL toward efficient reward and task generalization.
  arXiv  Detail & Related papers  (2025-06-02T04:21:56Z)
• ELITE: Embedding-Less retrieval with Iterative Text Exploration [5.8851517822935335]
  Large Language Models (LLMs) have achieved impressive progress in natural language processing.<n>Their limited ability to retain long-term context constrains performance on document-level or multi-turn tasks.
  arXiv  Detail & Related papers  (2025-05-17T08:48:43Z)
• PICASO: Permutation-Invariant Context Composition with State Space Models [98.91198288025117]
  State Space Models (SSMs) offer a promising solution by allowing a database of contexts to be mapped onto fixed-dimensional states.<n>We propose a simple mathematical relation derived from SSM dynamics to compose multiple states into one that efficiently approximates the effect of concatenating raw context tokens.<n>We evaluate our resulting method on WikiText and MSMARCO in both zero-shot and fine-tuned settings, and show that we can match the strongest performing baseline while enjoying on average 5.4x speedup.
  arXiv  Detail & Related papers  (2025-02-24T19:48:00Z)
• Core Context Aware Transformers for Long Context Language Modeling [50.774702091154204]
  We propose a plug-and-play Core Context Aware (CCA) Attention for efficient long-context modeling.<n>Our method automatically focuses and strengthens core context while diminishing redundancy during the learning process.<n>Our method is able to replace the self-attention module in existing Large Language Models with minimal fine-tuning cost.
  arXiv  Detail & Related papers  (2024-12-17T01:54:08Z)
• Recognizing Long Grammatical Sequences Using Recurrent Networks Augmented With An External Differentiable Stack [73.48927855855219]
  Recurrent neural networks (RNNs) are a widely used deep architecture for sequence modeling, generation, and prediction. RNNs generalize poorly over very long sequences, which limits their applicability to many important temporal processing and time series forecasting problems. One way to address these shortcomings is to couple an RNN with an external, differentiable memory structure, such as a stack. In this paper, we improve the memory-augmented RNN with important architectural and state updating mechanisms.
  arXiv  Detail & Related papers  (2020-04-04T14:19:15Z)

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.