# NLPerturbator: Studying the Robustness of Code LLMs to Natural Language Variations

arXiv: https://arxiv.org/abs/2406.19783 · HTML: https://mlsanity.com/paper/2406.19783/

## Key facts

- Cited by 20 papers in the map (data through 14 Oct 2026).
- Builds on 40 of its 68 references that are in the map.
- First posted on arXiv 28 Jun 2024.

## Builds on (most cited first, 30 of 40)

- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [Code Llama: Open Foundation Models for Code](https://mlsanity.com/paper/2308.12950/) (2023)
- [LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code](https://mlsanity.com/paper/2403.07974/) (2024)
- [CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation](https://mlsanity.com/paper/2109.00859/) (2021)
- [StarCoder: may the source be with you!](https://mlsanity.com/paper/2305.06161/) (2023)
- [WizardCoder: Empowering Code Large Language Models with Evol-Instruct](https://mlsanity.com/paper/2306.08568/) (2023)
- [InCoder: A Generative Model for Code Infilling and Synthesis](https://mlsanity.com/paper/2204.05999/) (2022)
- [CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning](https://mlsanity.com/paper/2207.01780/) (2022)
- [Synthetic and Natural Noise Both Break Neural Machine Translation](https://mlsanity.com/paper/1711.02173/) (2017)
- [CodeGeeX: A Pre-Trained Model for Code Generation with Multilingual Benchmarking on HumanEval-X](https://mlsanity.com/paper/2303.17568/) (2023)
- [Adversarial GLUE: A Multi-Task Benchmark for Robustness Evaluation of Language Models](https://mlsanity.com/paper/2111.02840/) (2021)
- [IntelliCode Compose: Code Generation Using Transformer](https://mlsanity.com/paper/2005.08025/) (2020)
- [PromptRobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Prompts](https://mlsanity.com/paper/2306.04528/) (2023)
- [On the Robustness of ChatGPT: An Adversarial and Out-of-distribution Perspective](https://mlsanity.com/paper/2302.12095/) (2023)
- [DocPrompting: Generating Code by Retrieving the Docs](https://mlsanity.com/paper/2207.05987/) (2022)
- [CodeGen2: Lessons for Training LLMs on Programming and Natural Languages](https://mlsanity.com/paper/2305.02309/) (2023)
- [ReCode: Robustness Evaluation of Code Generation Models](https://mlsanity.com/paper/2212.10264/) (2022)
- [Latent Predictor Networks for Code Generation](https://mlsanity.com/paper/1603.06744/) (2016)
- [Can ChatGPT replace StackOverflow? A Study on Robustness and Reliability of Large Language Model Code Generation](https://mlsanity.com/paper/2308.10335/) (2023)
- [Large Language Models are Few-Shot Summarizers: Multi-Intent Comment Generation via In-Context Learning](https://mlsanity.com/paper/2304.11384/) (2023)
- [On the Robustness of Code Generation Techniques: An Empirical Study on GitHub Copilot](https://mlsanity.com/paper/2302.00438/) (2023)
- [TreeGen: A Tree-Based Transformer Architecture for Code Generation](https://mlsanity.com/paper/1911.09983/) (2019)
- [Reasoning Runtime Behavior of a Program with LLM: How Far Are We?](https://mlsanity.com/paper/2403.16437/) (2024)
- [On Robustness of Prompt-based Semantic Parsing with Large Pre-trained Language Model: An Empirical Study on Codex](https://mlsanity.com/paper/2301.12868/) (2023)
- [In ChatGPT We Trust? Measuring and Characterizing the Reliability of ChatGPT](https://mlsanity.com/paper/2304.08979/) (2023)
- [Fill in the Blank: Context-aware Automated Text Input Generation for Mobile GUI Testing](https://mlsanity.com/paper/2212.04732/) (2022)
- [SOEN-101: Code Generation by Emulating Software Process Models Using Large Language Model Agents](https://mlsanity.com/paper/2403.15852/) (2024)

## Built on by (most cited first, 20 of 20)

- [CodeCrash: Exposing LLM Fragility to Misleading Natural Language in Code Reasoning](https://mlsanity.com/paper/2504.14119/) (2025)
- [Understanding the Effectiveness of Coverage Criteria for Large Language Models: A Special Angle from Jailbreak Attacks](https://mlsanity.com/paper/2408.15207/) (2024)
- [Memorize or Generalize? Evaluating LLM Code Generation with Code Rewriting](https://mlsanity.com/paper/2503.02296/) (2025)
- [RobuNFR: Evaluating the Robustness of Large Language Models on Non-Functional Requirements Aware Code Generation](https://mlsanity.com/paper/2503.22851/) (2025)
- [Evaluating Robustness of Large Language Models in Enterprise Applications: Benchmarks for Perturbation Consistency Across Formats and Languages](https://mlsanity.com/paper/2601.06341/) (2026)
- [Code Roulette: How Prompt Variability Affects LLM Code Generation](https://mlsanity.com/paper/2506.10204/) (2025)
- [Understanding and Bridging the Planner-Coder Gap: A Systematic Study on the Robustness of Multi-Agent Systems for Code Generation](https://mlsanity.com/paper/2510.10460/) (2025)
- [Defective Task Descriptions in LLM-Based Code Generation: Detection and Analysis](https://mlsanity.com/paper/2604.24703/) (2026)
- [Minimal Prompt Perturbations Lead to Code Vulnerabilities: Prompt Fragility and Hidden-State Signals in Coding LLMs](https://mlsanity.com/paper/2605.29737/) (2026)
- [A Representation Level Analysis of NMT Model Robustness to Grammatical Errors](https://mlsanity.com/paper/2505.21224/) (2025)
- [Prompt Stability in Code LLMs: Measuring Sensitivity across Emotion- and Personality-Driven Variations](https://mlsanity.com/paper/2509.13680/) (2025)
- [LLM Assisted Coding with Metamorphic Specification Mutation Agent](https://mlsanity.com/paper/2511.18249/) (2025)
- [Using Small Language Models to Reverse-Engineer Machine Learning Pipelines Structures](https://mlsanity.com/paper/2601.03988/) (2026)
- [Characterizing Paraphrase-Induced Failures in Lean 4 Autoformalization](https://mlsanity.com/paper/2604.23135/) (2026)
- [Detecting Differences Is Not Understanding Structure: Large Language Models Fail at Graph Isomorphism](https://mlsanity.com/paper/2606.09484/) (2026)
- [SWE-NFI: Studying and Benchmarking Coding Agents for Non-Functional Improvements](https://mlsanity.com/paper/2607.27409/) (2026)
- [ML For Hardware Design Interpretability: Challenges and Opportunities](https://mlsanity.com/paper/2504.08852/) (2025)
- [Memorization Diagnostics for Code LLMs Should be Scale-Aware](https://mlsanity.com/paper/2608.12771/) (2026)
- [RealSWE: A Compositional Evaluation of Coding Agents under Realistic User Requests](https://mlsanity.com/paper/2608.27831/) (2026)
- [Noisy Test-Time Reinforcement Learning for Code LLMs](https://mlsanity.com/paper/2609.32172/) (2026)

## Abstract (arXiv)

Large language models (LLMs) achieve promising results in code generation based on a given natural language description. They have been integrated into open-source projects and commercial products to facilitate daily coding activities. The natural language description in the prompt is crucial for LLMs to comprehend users' requirements. Prior studies uncover that LLMs are sensitive to the changes in the prompts, including slight changes that look inconspicuous. However, the natural language descriptions often vary in real-world scenarios (e.g., different formats, grammar, and wording). Prior studies on the robustness of LLMs are often based on random perturbations and such perturbations may not actually happen. In this paper, we conduct a comprehensive study to investigate how are code LLMs robust to variations of natural language description in real-world scenarios. We summarize 18 categories of perturbations of natural language and 3 combinations of co-occurred categories based on our literature review and an online survey with practitioners. We propose an automated framework, NLPerturbator, which can perform perturbations of each category given a set of prompts. Through a series of experiments on code generation using six code LLMs, we find that the perturbed prompts can decrease the performance of code generation by a considerable margin (e.g., up to 21.2%, and 4.8% to 6.1% on average). Our study highlights the importance of enhancing the robustness of LLMs to real-world variations in the prompts, as well as the essentiality of attentively constructing the prompts.