Language Models (Mostly) Know When to Stop Reading

• URL: http://arxiv.org/abs/2502.01025v2
• Date: Wed, 22 Oct 2025 21:46:56 GMT
• Title: Language Models (Mostly) Know When to Stop Reading
• Authors: Roy Xie, Junlin Wang, Paul Rosu, Chunyuan Deng, Bolun Sun, Zihao Lin, Bhuwan Dhingra,
• Abstract summary: Large language models (LLMs) process entire input contexts indiscriminately, which is inefficient when the information required to answer a query is localized within the context.<n>We present dynamic context cutoff, a novel method enabling LLMs to self-terminate processing upon acquiring sufficient task-relevant information.
• Score: 24.246459354913146
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) process entire input contexts indiscriminately, which is inefficient when the information required to answer a query is localized within the context. We present dynamic context cutoff, a novel method enabling LLMs to self-terminate processing upon acquiring sufficient task-relevant information. Through analysis of model internals, we discover that specific attention heads inherently encode "sufficiency signals" -- detectable through lightweight classifiers -- that predict when critical information has been processed. This reveals a new efficiency paradigm: models' internal understanding naturally dictates processing needs rather than external compression heuristics. Comprehensive experiments across six QA datasets (up to 40K tokens) with three model families (LLaMA/Qwen/Mistral, 1B-70B) demonstrate 3.4% accuracy improvement while achieving 1.33x token reduction on average. Furthermore, our method demonstrates superior performance compared to other context efficiency methods at equivalent token reduction rates. Additionally, we observe an emergent scaling phenomenon: while smaller models require probing for sufficiency detection, larger models exhibit intrinsic self-assessment capabilities through prompting.

Related papers

• Information Capacity: Evaluating the Efficiency of Large Language Models via Text Compression [53.39128997308138]
  We introduce information capacity, a measure of model efficiency based on text compression performance.<n> Empirical evaluations on mainstream open-source models show that models of varying sizes within a series exhibit consistent information capacity.<n>A distinctive feature of information capacity is that it incorporates tokenizer efficiency, which affects both input and output token counts.
  arXiv  Detail & Related papers  (2025-11-11T10:07:32Z)
• LLM Microscope: What Model Internals Reveal About Answer Correctness and Context Utilization [9.410181019585822]
  We operationalize interpretability methods to ascertain whether we can predict the correctness of model outputs.<n>We consider correct, incorrect, and irrelevant context and introduce metrics to distinguish amongst them.<n>Our model-internals-based metric significantly outperforms prompting baselines at distinguishing between correct and incorrect context.
  arXiv  Detail & Related papers  (2025-10-05T03:14:05Z)
• SPaRFT: Self-Paced Reinforcement Fine-Tuning for Large Language Models [51.74498855100541]
  Large language models (LLMs) have shown strong reasoning capabilities when fine-tuned with reinforcement learning (RL)<n>We propose textbfSPaRFT, a self-paced learning framework that enables efficient learning based on the capability of the model being trained.
  arXiv  Detail & Related papers  (2025-08-07T03:50:48Z)
• Preemptive Hallucination Reduction: An Input-Level Approach for Multimodal Language Model [1.124958340749622]
  This study presents a novel ensemble-based preprocessing framework that adaptively selects the most appropriate filtering approach.<n>The method achieves a 44.3% reduction in hallucination rates, as measured by Natural Language Inference (NLI) scores.<n>The findings highlight the importance of adaptive preprocessing techniques in mitigating hallucinations, paving the way for more reliable multimodal systems.
  arXiv  Detail & Related papers  (2025-05-29T21:09:34Z)
• Self-Route: Automatic Mode Switching via Capability Estimation for Efficient Reasoning [36.470695895695044]
  Self-Route is a dynamic reasoning framework that automatically selects between general and reasoning modes.<n>We show that Self-Route achieves comparable accuracy to reasoning models while reducing token consumption by 30-55%.
  arXiv  Detail & Related papers  (2025-05-27T03:18:31Z)
• Learning Task Representations from In-Context Learning [73.72066284711462]
  Large language models (LLMs) have demonstrated remarkable proficiency in in-context learning. We introduce an automated formulation for encoding task information in ICL prompts as a function of attention heads. We show that our method's effectiveness stems from aligning the distribution of the last hidden state with that of an optimally performing in-context-learned model.
  arXiv  Detail & Related papers  (2025-02-08T00:16:44Z)
• 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)
• Likelihood as a Performance Gauge for Retrieval-Augmented Generation [78.28197013467157]
  We show that likelihoods serve as an effective gauge for language model performance. We propose two methods that use question likelihood as a gauge for selecting and constructing prompts that lead to better performance.
  arXiv  Detail & Related papers  (2024-11-12T13:14:09Z)
• M$^3$-Impute: Mask-guided Representation Learning for Missing Value Imputation [12.174699459648842]
  M$3$-Impute aims to explicitly leverage the missingness information and such correlations with novel masking schemes. Experiment results show the effectiveness of M$3$-Impute by achieving 20 best and 4 second-best MAE scores on average.
  arXiv  Detail & Related papers  (2024-10-11T13:25:32Z)
• TernaryLLM: Ternarized Large Language Model [29.29122031050894]
  Large language models (LLMs) have achieved remarkable performance on Natural Language Processing (NLP) tasks. We introduce Dual Learnable Ternarization (DLT), which enables both scales and shifts to be learnable. We also propose Outlier-Friendly Feature Knowledge Distillation (OFF) to recover the information lost in extremely low-bit quantization.
  arXiv  Detail & Related papers  (2024-06-11T11:40:12Z)
• Unsupervised Pre-training with Language-Vision Prompts for Low-Data Instance Segmentation [105.23631749213729]
  We propose a novel method for unsupervised pre-training in low-data regimes. Inspired by the recently successful prompting technique, we introduce a new method, Unsupervised Pre-training with Language-Vision Prompts. We show that our method can converge faster and perform better than CNN-based models in low-data regimes.
  arXiv  Detail & Related papers  (2024-05-22T06:48:43Z)
• In-Context Language Learning: Architectures and Algorithms [73.93205821154605]
  We study ICL through the lens of a new family of model problems we term in context language learning (ICLL) We evaluate a diverse set of neural sequence models on regular ICLL tasks.
  arXiv  Detail & Related papers  (2024-01-23T18:59:21Z)
• Making Retrieval-Augmented Language Models Robust to Irrelevant Context [55.564789967211844]
  An important desideratum of RALMs, is that retrieved information helps model performance when it is relevant. Recent work has shown that retrieval augmentation can sometimes have a negative effect on performance.
  arXiv  Detail & Related papers  (2023-10-02T18:52:35Z)
• The Devil is in the Errors: Leveraging Large Language Models for Fine-grained Machine Translation Evaluation [93.01964988474755]
  AutoMQM is a prompting technique which asks large language models to identify and categorize errors in translations. We study the impact of labeled data through in-context learning and finetuning. We then evaluate AutoMQM with PaLM-2 models, and we find that it improves performance compared to just prompting for scores.
  arXiv  Detail & Related papers  (2023-08-14T17:17:21Z)
• Information Association for Language Model Updating by Mitigating LM-Logical Discrepancy [68.31760483418901]
  Large Language Models(LLMs) struggle with providing current information due to the outdated pre-training data. Existing methods for updating LLMs, such as knowledge editing and continual fine-tuning, have significant drawbacks in generalizability of new information. We identify the core challenge behind these drawbacks: the LM-logical discrepancy featuring the difference between language modeling probabilities and logical probabilities.
  arXiv  Detail & Related papers  (2023-05-29T19:48:37Z)
• Dynamic Context Pruning for Efficient and Interpretable Autoregressive Transformers [29.319666323947708]
  We present a novel approach that dynamically prunes contextual information while preserving the model's expressiveness. Our method employs a learnable mechanism that determines which uninformative tokens can be dropped from the context. Our reference implementation achieves up to $2times$ increase in inference throughput and even greater memory savings.
  arXiv  Detail & Related papers  (2023-05-25T07:39:41Z)
• Discover, Explanation, Improvement: An Automatic Slice Detection Framework for Natural Language Processing [72.14557106085284]
  slice detection models (SDM) automatically identify underperforming groups of datapoints. This paper proposes a benchmark named "Discover, Explain, improve (DEIM)" for classification NLP tasks. Our evaluation shows that Edisa can accurately select error-prone datapoints with informative semantic features.
  arXiv  Detail & Related papers  (2022-11-08T19:00:00Z)
• Inference from Real-World Sparse Measurements [21.194357028394226]
  Real-world problems often involve complex and unstructured sets of measurements, which occur when sensors are sparsely placed in either space or time. Deep learning architectures capable of processing sets of measurements with positions varying from set to set and extracting readouts anywhere are methodologically difficult. We propose an attention-based model focused on applicability and practical robustness, with two key design contributions.
  arXiv  Detail & Related papers  (2022-10-20T13:42:20Z)
• Correlation Information Bottleneck: Towards Adapting Pretrained Multimodal Models for Robust Visual Question Answering [63.87200781247364]
  Correlation Information Bottleneck (CIB) seeks a tradeoff between compression and redundancy in representations. We derive a tight theoretical upper bound for the mutual information between multimodal inputs and representations.
  arXiv  Detail & Related papers  (2022-09-14T22:04:10Z)
• Layer-wise Analysis of a Self-supervised Speech Representation Model [26.727775920272205]
  Self-supervised learning approaches have been successful for pre-training speech representation models. Not much has been studied about the type or extent of information encoded in the pre-trained representations themselves.
  arXiv  Detail & Related papers  (2021-07-10T02:13:25Z)

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.