# EyeLayer: Integrating Human Attention Patterns into LLM-Based Code Summarization

arXiv: https://arxiv.org/abs/2602.22368 · HTML: https://mlsanity.com/paper/2602.22368/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 51 references that are in the map.
- First posted on arXiv 25 Feb 2026.

## Builds on (most cited first, 28 of 28)

- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [CodeBERT: A Pre-Trained Model for Programming and Natural Languages](https://mlsanity.com/paper/2002.08155/) (2020)
- [Gradient Surgery for Multi-Task Learning](https://mlsanity.com/paper/2001.06782/) (2020)
- [CodeXGLUE: A Machine Learning Benchmark Dataset for Code Understanding and Generation](https://mlsanity.com/paper/2102.04664/) (2021)
- [CodeSearchNet Challenge: Evaluating the State of Semantic Code Search](https://mlsanity.com/paper/1909.09436/) (2019)
- [Generating Sequences With Recurrent Neural Networks](https://mlsanity.com/paper/1308.0850/) (2013)
- [Attention-Based Models for Speech Recognition](https://mlsanity.com/paper/1506.07503/) (2015)
- [On the Relationship between Self-Attention and Convolutional Layers](https://mlsanity.com/paper/1911.03584/) (2019)
- [Grounded Copilot: How Programmers Interact with Code-Generating Models](https://mlsanity.com/paper/2206.15000/) (2022)
- [Do Users Write More Insecure Code with AI Assistants?](https://mlsanity.com/paper/2211.03622/) (2022)
- [DRAW: A Recurrent Neural Network For Image Generation](https://mlsanity.com/paper/1502.04623/) (2015)
- [Large Language Models for Software Engineering: Survey and Open Problems](https://mlsanity.com/paper/2310.03533/) (2023)
- [code2seq: Generating Sequences from Structured Representations of Code](https://mlsanity.com/paper/1808.01400/) (2018)
- [A Transformer-based Approach for Source Code Summarization](https://mlsanity.com/paper/2005.00653/) (2020)
- [On the Evaluation of Neural Code Summarization](https://mlsanity.com/paper/2107.07112/) (2021)
- [What Makes Good In-context Demonstrations for Code Intelligence Tasks with LLMs?](https://mlsanity.com/paper/2304.07575/) (2023)
- [Hard-Coded Gaussian Attention for Neural Machine Translation](https://mlsanity.com/paper/2005.00742/) (2020)
- [Talking Heads: Understanding Inter-layer Communication in Transformer Language Models](https://mlsanity.com/paper/2406.09519/) (2024)
- [Security Weaknesses of Copilot-Generated Code in GitHub Projects: An Empirical Study](https://mlsanity.com/paper/2310.02059/) (2023)
- [Investigating Layer Importance in Large Language Models](https://mlsanity.com/paper/2409.14381/) (2024)
- [Distilled GPT for Source Code Summarization](https://mlsanity.com/paper/2308.14731/) (2023)
- [Evaluating Code Summarization Techniques: A New Metric and an Empirical Characterization](https://mlsanity.com/paper/2312.15475/) (2023)
- [Do Machines and Humans Focus on Similar Code? Exploring Explainability of Large Language Models in Code Summarization](https://mlsanity.com/paper/2402.14182/) (2024)
- [Calibration of Large Language Models on Code Summarization](https://mlsanity.com/paper/2404.19318/) (2024)
- [EyeTrans: Merging Human and Machine Attention for Neural Code Summarization](https://mlsanity.com/paper/2402.14096/) (2024)
- [Towards Modeling Human Attention from Eye Movements for Neural Source Code Summarization](https://mlsanity.com/paper/2305.09773/) (2023)
- [EyeMulator: Improving Code Language Models by Mimicking Human Visual Attention](https://mlsanity.com/paper/2508.16771/) (2025)
- [Disentangling Recall and Reasoning in Transformer Models through Layer-wise Attention and Activation Analysis](https://mlsanity.com/paper/2510.03366/) (2025)

## Abstract (arXiv)

Code summarization is the task of generating natural language descriptions of source code, which is critical for software comprehension and maintenance. While large language models (LLMs) have achieved remarkable progress on this task, an open question remains: can human expertise in code understanding further guide and enhance these models? We propose EyeLayer, a lightweight attention-augmentation module that incorporates human eye-gaze patterns, as a proxy of human expertise, into LLM-based code summarization. EyeLayer models human attention during code reading via a Multimodal Gaussian Mixture, redistributing token embeddings based on learned parameters (μi, σi2) that capture where and how intensively developers focus. This design enables learning generalizable attention priors from eye-tracking data and incorporating them into LLMs seamlessly, without disturbing existing representations. We evaluate EyeLayer across diverse model families (i.e., LLaMA-3.2, Qwen3, and CodeBERT) covering different scales and architectures. EyeLayer consistently outperforms strong fine-tuning baselines across standard metrics, achieving gains of up to 13.17% on BLEU-4. These results demonstrate that human gaze patterns encode complementary attention signals that enhance the semantic focus of LLMs and transfer effectively across diverse models for code summarization.