Don't Complete It! Preventing Unhelpful Code Completion for Productive and Sustainable Neural Code Completion Systems
Abstract
Currently, large pre-trained language models are widely applied in neural code completion systems. Though large code models significantly outperform their smaller counterparts, around 70% of displayed code completions from Copilot are not accepted by developers. Being reviewed but not accepted, their help to developer productivity is considerably limited. Even worse, considering the high cost of the large code models, it is a huge waste of computing resources and energy. To fill this significant gap, we propose an early-rejection mechanism to turn down low-return prompts by foretelling the code completion qualities without sending them to the code completion system. Furthermore, we propose a lightweight Transformer-based es-timator to demonstrate the feasibility of the mechanism. The experimental results show that the proposed estimator helps save 23.3% of computational cost measured in floating-point operations for the code completion systems, and 80.2% of rejected prompts lead to unhelpful completion.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Sun et al. assume completions are rejected based on the context before the cursor alone, ignoring the interplay with developers' mode of thought.
“However, Sun et al. [39] assume completions are rejected based on the context before the cursor alone, ignoring the interplay with developers' mode of thought.”
From A Transformer-Based Approach for Smart Invocation of Automatic Code Completion (2024) · cites this paper as Sun et al. [39]
Both Sun et al. and Mozannar et al. aim to maximise the acceptance rate of the shown completions.
“Moreover, both of these tools aim to maximise the acceptance rate of the shown completions”
From A Transformer-Based Approach for Smart Invocation of Automatic Code Completion (2024) · cites this paper as [39]
Sun et al.'s transformer-based classifier can hide 5% of suggestions that would have been rejected with 94.5% accuracy.
“They train a transformer-based classifier on the code prefix before it is submitted to the completion model, and find it can hide 5% of suggestions that would have been rejected with 94.5% accuracy.”
From A Transformer-Based Approach for Smart Invocation of Automatic Code Completion (2024) · cites this paper as Sun et al. [39]
Cites
Productivity Assessment of Neural Code Completion
Ziegler, A., Kalliamvakou, E., Li, X. A., Rice, A., Rifkin, D., Simister, S., Sittampalam, G., & Aftandilian, E. (2022) · Proceedings of the 6th ACM SIGPLAN International Symposium on Machine Programming · doi:10.1145/3520312.3534864 · Cited by 200 · see citing works
Productivity Assessment of Neural Code Completion
Ziegler, A., Kalliamvakou, E., Simister, S., Sittampalam, G., Li, A., Rice, A., Rifkin, D., & Aftandilian, E. (2022) · arXiv · doi:10.48550/arxiv.2205.06537 · Cited by 200 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Reading Between the Lines: Modeling User Behavior and Costs in AI-Assisted Programming
Mozannar, H., Bansal, G., Fourney, A., & Horvitz, E. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3641936 · Cited by 114 · see citing works
Generative AI in mobile networks: a survey
Karapantelakis, A., Alizadeh, P., Alabbasi, A., Dey, K., & Nikou, A. (2023) · Annals of Telecommunications · doi:10.1007/s12243-023-00980-9 · Cited by 85 · see citing works
Shi, J., Yang, Z., & Lo, D. (2025) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3708525 · Cited by 43 · see citing works
Multi-line AI-Assisted Code Authoring
Dunay, O., Cheng, D., Tait, A., Thakkar, P., Rigby, P. C., Chiu, A., Ahmad, I., Ganesan, A., Maddila, C., Murali, V., et al. (2024) · SIGSOFT FSE Companion · doi:10.1145/3663529.3663836 · Cited by 29 · see citing works
Beyond Functional Correctness: Investigating Coding Style Inconsistencies in Large Language Models
Wang, Y., Jiang, T., Liu, M., Chen, J., Mao, M., Liu, X., Ma, Y., & Zheng, Z. (2025) · Proceedings of the ACM on software engineering. · doi:10.1145/3715749 · Cited by 21 · see citing works
"My productivity is boosted, but …" Demystifying Users’ Perception on AI Coding Assistants
Lyu, Y., Yang, Z., Shi, J., Chang, J., Liu, Y., & Lo, D. (2025) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1109/ase63991.2025.00024 · Cited by 9 · see citing works
A Transformer-Based Approach for Smart Invocation of Automatic Code Completion
de Moor, A., van Deursen, A., & Izadi, M. (2024) · Proceedings of the 1st ACM International Conference on AI-Powered Software · doi:10.1145/3664646.3664760 · Cited by 7 · see citing works
Generating and Reviewing Programming Codes with Large Language Models: A Systematic Mapping Study
de Albuquerque, B. V. L., da Cunha, A. F. S., Souza, L., Siqueira, S. W. M., & dos Santos, R. P. (2024) · Anais do Simpósio Brasileiro de Sistemas de Informação (SBSI) · doi:10.1145/3658271.3658342 · Cited by 4 · see citing works
How do Copilot Suggestions Impact Developers' Frustration and Productivity?
Guglielmi, E., Arnoudova, V., Bavota, G., Oliveto, R., & Scalabrino, S. (2025) · arXiv · doi:10.48550/arxiv.2504.06808 · Cited by 1 · see citing works
A Transformer-Based Approach for Smart Invocation of Automatic Code Completion
de Moor, A., van Deursen, A., & Izadi, M. (2024) · arXiv
- Authors
- Sun, Z., Du, X., Song, F., Wang, S., Ni, M., & Li, L.
- Affiliations
- ShanghaiTech University (CN)
- Monash University (AU)
- National University of Defense Technology (CN)
- University of Technology Sydney (AU)
- Venue
- 2023 IEEE/ACM 45th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion)
- Publisher
- IEEE
- Published
- 2023-05
- Pages
- 324-325
- ISSN
- 0270-5257 (ISSN-L)
- OpenAlex
- W4384026646
- Cited by
- 8
- References
- 1
- Citations by year
- 2025: 2 · 2024: 5 · 2023: 1
- FWCI
- 2.05
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM/IEEE International Conference on Software Engineering: Companion Proceedings (ICSE-Companion) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Information Systems › Software Engineering Research
- Physical Sciences › Computer Science › Information Systems › Software Engineering Techniques and Practices
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- early rejection, large language models, Copilot, prompt filtering, computational complexity reduction, energy efficiency, developer productivity
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
