PulseAugur
EN
LIVE 17:01:23

LLMs improve code generation with iterative feedback and efficiency focus

By PulseAugur Editorial · [7 sources] ·

Researchers are exploring new methods to improve the reliability and efficiency of code generated by large language models (LLMs). One study focuses on iterative feedback loops, using compiler error messages and test case results to help LLMs refine their code, with reasoning models showing significant improvement over non-reasoning ones. Another paper introduces SwiftTrans, a framework designed to enhance both the functional correctness and runtime efficiency of LLM-translated code, addressing the issue that LLM-generated programs can be slower than human-written equivalents. A third approach tackles code hallucination by selectively generating code and using automatically generated unit tests to evaluate and control correctness. AI

IMPACT These research papers explore methods to enhance the accuracy, efficiency, and reliability of LLM-generated code, potentially accelerating adoption in safety-critical applications.

RANK_REASON Multiple research papers published on arXiv detailing new methods for improving LLM code generation capabilities.

Read on Hugging Face Daily Papers →

AI-generated summary · Google Gemini · from 7 sources. How we write summaries →

LLMs improve code generation with iterative feedback and efficiency focus

COVERAGE [7]

  1. arXiv cs.AI TIER_1 English(EN) · Zepeng Li, Jie Ren, Zhanyong Tang, Jie Zheng, Zheng Wang ·

    AutoPass: Evidence-Guided LLM Agents for Compiler Performance Tuning

    arXiv:2606.20373v1 Announce Type: cross Abstract: Large Language Models (LLMs) show promise for code compilation tasks, but applying them to runtime performance tuning is difficult due to complex microarchitectural effects and noisy runtime measurements. We present AutoPass, a mu…

  2. arXiv cs.AI TIER_1 English(EN) · Zheng Wang ·

    AutoPass: Evidence-Guided LLM Agents for Compiler Performance Tuning

    Large Language Models (LLMs) show promise for code compilation tasks, but applying them to runtime performance tuning is difficult due to complex microarchitectural effects and noisy runtime measurements. We present AutoPass, a multi-agent framework for compiler performance tunin…

  3. arXiv cs.CL TIER_1 English(EN) · Longhui Zhang, Jiahao Wang, Chenhao Hu, Bingyu Liang, Jing Li, Min Zhang ·

    Bridging Functional Correctness and Runtime Efficiency Gaps in LLM-Based Code Translation

    arXiv:2606.17683v1 Announce Type: new Abstract: While large language models (LLMs) have greatly advanced the functional correctness of automated code translation systems, the runtime efficiency of translated programs has received comparatively little attention. With the waning of…

  4. arXiv cs.AI TIER_1 English(EN) · Le Zhang, Suresh Kothari ·

    Unlocking LLM Code Correction with Iterative Feedback Loops

    arXiv:2606.17514v1 Announce Type: cross Abstract: Large Language Models have shown remarkable capabilities in code generation. However, most existing evaluations focus only on single-attempt accuracy and overlook the iterative refinement process that is central to real-world prog…

  5. arXiv cs.CL TIER_1 English(EN) · Min Zhang ·

    Bridging Functional Correctness and Runtime Efficiency Gaps in LLM-Based Code Translation

    While large language models (LLMs) have greatly advanced the functional correctness of automated code translation systems, the runtime efficiency of translated programs has received comparatively little attention. With the waning of Moore's law, runtime efficiency has become incr…

  6. arXiv cs.LG TIER_1 English(EN) · Jaewoo Jeong, Taesoo Kim, Sangdon Park ·

    Towards Functional Correctness of Large Code Models with Selective Generation

    arXiv:2505.13553v3 Announce Type: replace-cross Abstract: The hallucination of code generation models hinders their applicability to systems requiring higher safety standards. One critical bottleneck in addressing code hallucination is the difficulty of identifying the functional…

  7. Hugging Face Daily Papers TIER_1 English(EN) ·

    No Resource, No Benchmarks, No Problem? Evaluating and Improving LLMs for Code Generation in No-Resource Languages

    Research addresses code generation challenges for no-resource programming languages by developing benchmarks and proposing a method that combines further pre-training with weight difference transfer to create specialized instruction-following models at reduced computational cost.

RELATED ENTITIES

RELATED TOPICS