EyeMulator: Improving Code Language Models by Mimicking Human Visual Attention

• URL: http://arxiv.org/abs/2508.16771v1
• Date: Fri, 22 Aug 2025 20:08:09 GMT
• Title: EyeMulator: Improving Code Language Models by Mimicking Human Visual Attention
• Abstract summary: We present EyeMulator, a technique for training CodeLLMs to mimic human visual attention while training for various software development tasks.<n>We draw these weights from observations of human visual attention derived from a previously-collected publicly-available dataset of eye-tracking experiments in software engineering tasks.<n>Our evaluation shows that EyeMulator outperforms strong LLM baselines on several tasks such as code translation, completion and summarization.
• Score: 27.11897727181663
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Code language models (so-called CodeLLMs) are now commonplace in software development. As a general rule, CodeLLMs are trained by dividing training examples into input tokens and then learn importance of those tokens in a process called machine attention. Machine attention is based solely on input token salience to output token examples during training. Human software developers are different, as humans intuitively know that some tokens are more salient than others. While intuition itself is ineffable and a subject of philosophy, clues about salience are present in human visual attention, since people tend to look at more salient words more often. In this paper, we present EyeMulator, a technique for training CodeLLMs to mimic human visual attention while training for various software development tasks. We add special weights for each token in each input example to the loss function used during LLM fine-tuning. We draw these weights from observations of human visual attention derived from a previously-collected publicly-available dataset of eye-tracking experiments in software engineering tasks. These new weights ultimately induce changes in the attention of the subject LLM during training, resulting in a model that does not need eye-tracking data during inference. Our evaluation shows that EyeMulator outperforms strong LLM baselines on several tasks such as code translation, completion and summarization. We further show an ablation study that demonstrates the improvement is due to subject models learning to mimic human attention.

Related papers

• Predicting Human Visual Attention on Words in Source Code [0.7448254811651417]
  We design a novel loss function that computes similarity between human attention measured during eye tracking experiments and the internal attention of the artificial neural network.<n>Our model achieves a statistically significant improvement over the close baseline in the reading task according to normalized Levenshtein distance.
  arXiv  Detail & Related papers  (2026-07-16T03:42:16Z)
• EyeLayer: Integrating Human Attention Patterns into LLM-Based Code Summarization [15.602166020026045]
  EyeLayer is a lightweight attention-augmentation module that incorporates human eye-gaze patterns.<n>We show that EyeLayer consistently outperforms strong fine-tuning baselines across standard metrics.
  arXiv  Detail & Related papers  (2026-02-25T20:04:39Z)
• Learning to See Before Seeing: Demystifying LLM Visual Priors from Language Pre-training [37.93241751782069]
  Large Language Models (LLMs) develop rich visual priors despite being trained on text alone.<n>These priors allow latent visual capabilities to be unlocked for vision tasks with a relatively small amount of multimodal data.<n>We show that visual priors are composed of separable perception and reasoning priors with unique scaling trends and origins.
  arXiv  Detail & Related papers  (2025-09-30T17:57:44Z)
• Visual Jigsaw Post-Training Improves MLLMs [58.29961336087896]
  We introduce Visual Jigsaw, a generic self-supervised post-training framework designed to strengthen visual understanding in large language models (MLLMs)<n>Visual Jigsaw is formulated as a general ordering task: visual inputs are partitioned, shuffled, and the model must reconstruct the visual information by producing the correct permutation in natural language.<n>Extensive experiments demonstrate substantial improvements in fine-grained perception, temporal reasoning, and 3D spatial understanding.
  arXiv  Detail & Related papers  (2025-09-29T17:59:57Z)
• Introducing Visual Perception Token into Multimodal Large Language Model [53.82301522384719]
  Multimodal Large Language Model (MLLM) relies on the perception process of its vision encoder.<n>MLLM still lacks the autonomous capability to control its own visual perception processes.<n>We propose the concept of Visual Perception Token, aiming to empower MLLM with a mechanism to control its visual perception processes.
  arXiv  Detail & Related papers  (2025-02-24T18:56:12Z)
• Open Eyes, Then Reason: Fine-grained Visual Mathematical Understanding in MLLMs [62.875934732547435]
  Current large language models (MLLMs) often underperform on mathematical problem-solving tasks that require fine-grained visual understanding.<n>In this paper, we evaluate the visual grounding capabilities of state-of-the-art MLLMs and reveal a significant negative correlation between visual grounding accuracy and problem-solving performance.<n>We propose a novel approach, SVE-Math, featuring a geometric-grounded vision encoder and a feature router that dynamically adjusts the contribution of hierarchical visual feature maps.
  arXiv  Detail & Related papers  (2025-01-11T04:08:44Z)
• [CLS] Token Tells Everything Needed for Training-free Efficient MLLMs [66.5266435598799]
  Multi-language Large Language Models (MLLMs) have recently demonstrated strong performance across a wide range of vision tasks.<n>However, their efficient deployment remains a substantial challenge due to high computational costs and memory requirements.<n>We introduce a simple yet effective method for train-free visual compression, called VTC- compression.
  arXiv  Detail & Related papers  (2024-12-08T05:29:39Z)
• Looking Beyond Text: Reducing Language bias in Large Vision-Language Models via Multimodal Dual-Attention and Soft-Image Guidance [67.26434607115392]
  Large vision-language models (LVLMs) have achieved impressive results in various vision-language tasks. LVLMs suffer from hallucinations caused by language bias, leading to diminished focus on images and ineffective visual comprehension. We propose LACING to address the language bias of LVLMs with muLtimodal duAl-attention meChanIsm (MDA) aNd soft-image Guidance (IFG)
  arXiv  Detail & Related papers  (2024-11-21T16:33:30Z)
• ControlMLLM: Training-Free Visual Prompt Learning for Multimodal Large Language Models [73.34709921061928]
  We propose a training-free method to inject visual prompts into Multimodal Large Language Models (MLLMs)<n>We optimize a learnable latent variable based on an energy function, enhancing the strength of referring regions in the attention map.<n>Our method offers a promising direction for integrating referring abilities into MLLMs, and supports referring with box, mask, scribble and point.
  arXiv  Detail & Related papers  (2024-07-31T11:40:29Z)
• What is the Visual Cognition Gap between Humans and Multimodal LLMs? [63.81347276258992]
  We evaluate the visual cognition capability of Multimodal Large Language Models (MLLMs) and compare their performance with human visual cognition studies.<n>Our comparative experiments with different baselines reveal a gap between MLLMs and human intelligence.<n>We believe that the public release of MaRs-VQA and the Qwen2-VCog baseline model will drive progress toward the next generation of MLLMs with human-like visual cognition abilities.
  arXiv  Detail & Related papers  (2024-06-14T22:02:21Z)
• Naturalness of Attention: Revisiting Attention in Code Language Models [3.756550107432323]
  Language models for code such as CodeBERT offer the capability to learn advanced source code representation, but their opacity poses barriers to understanding of captured properties. This study aims to shed some light on the previously ignored factors of the attention mechanism beyond the attention weights.
  arXiv  Detail & Related papers  (2023-11-22T16:34:12Z)
• Multimodality and Attention Increase Alignment in Natural Language Prediction Between Humans and Computational Models [0.8139163264824348]
  Humans are known to use salient multimodal features, such as visual cues, to facilitate the processing of upcoming words. multimodal computational models can integrate visual and linguistic data using a visual attention mechanism to assign next-word probabilities. We show that predictability estimates from humans aligned more closely with scores generated from multimodal models vs. their unimodal counterparts.
  arXiv  Detail & Related papers  (2023-08-11T09:30:07Z)
• Decoding Attention from Gaze: A Benchmark Dataset and End-to-End Models [6.642042615005632]
  Eye-tracking has potential to provide rich behavioral data about human cognition in ecologically valid environments. This paper studies using computer vision tools for "attention decoding", the task of assessing the locus of a participant's overt visual attention over time.
  arXiv  Detail & Related papers  (2022-11-20T12:24:57Z)
• Follow-up Attention: An Empirical Study of Developer and Neural Model Code Exploration [6.060235526273212]
  Recent neural models of code, such as OpenAI Codex and AlphaCode, have demonstrated remarkable proficiency at code generation. However, it often remains unclear how the models actually process code, and to what extent their reasoning and the way their attention mechanism scans the code matches the patterns of developers. This work studies how the processed attention signal of three open large language models - CodeGen, InCoder and GPT-J - agrees with how developers look at and explore code.
  arXiv  Detail & Related papers  (2022-10-11T14:58:58Z)
• Brief Introduction to Contrastive Learning Pretext Tasks for Visual Representation [0.0]
  We introduce contrastive learning, a subset of unsupervised learning methods. The purpose of contrastive learning is to embed augmented samples from the same sample near to each other while pushing away those that are not. We offer some strategies from contrastive learning that have recently been published and are focused on pretext tasks for visual representation.
  arXiv  Detail & Related papers  (2022-10-06T18:54:10Z)

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.