AgentFlare

Scholar · code generation large language models

Code Generation with LLMs: Research Digest

10 foundational papers with 7,015 combined citations, plus the latest arXiv preprints.

Foundational papers107,015 citations
Most citedIs your code generated by chatgpt rea…2,721 citations · 2023
arXiv, Sep 202685 papers-31% vs Oct 2025
Newest preprint2026-10-15

AISA Scholar API · arXiv

arXiv submissions per monthPapers matching abs:"code generation" AND abs:"language model" · last 12 complete months
monthly arXiv submissions: Oct 2025 123 → Sep 2026 85; high 136 (Apr 2026), low 72 (Jul 2026)6792116141Oct 2025Mar 2026Sep 2026
402 · Pay-per-Crawl$0.02

You're reading the full report — free for humans and search engines. The same content ships as a structured JSON dataset (normalized facts, sources and citation arrays, no HTML parsing) so AI agents can consume it directly — metered per fetch via the open HTTP 402 payment protocol.

⚡ Agents: fetch .data.json · $0.02 → agents: crawlerAutoPrice=true · verify at https://esa.aisa.one/api/v1/access/verify

What’s new on arXiv

A12026-10-15cs.AI

RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent

Huan Zhang, Mingju Chen, Dongxu Zhou, Can Lv, Sen Cui, Faguo Wu, Shiji Zhou

Large language model (LLM) agents have demonstrated strong capabilities in complex tasks, including code generation, tool use, and multi-hop question answering. Existing self-improvement methods enhance agent performance through model optimization, harness evolution, and executable artifact refinement, while recursive self-improvement (RSI) further extends optimization to the improvement mechanisms themselves. However, adapting self-improvement workflows to evolving performance bottlenecks while improving workflow orchestration through execution feedback remains a key challenge. We propose RSIFlow, a Recursive Self-Improvement framework that integrates adaptive workflow orchestration with meta-agent capability evolution. RSIFlow diagnoses performance bottlenecks from task-agent trajectories and environmental feedback to dynamically select component-specific improvement workflows. The meta-agent then distills workflow outcomes into component-specific skills and cross-component principles, evolving its own harness to improve subsequent workflow orchestration. Experiments across six benchmarks and two backbone models show that RSIFlow achieves the SOTA average performance on all benchmarks. Ablation studies further demonstrate the contributions of adaptive workflow orchestration and meta-harness evolution. Code is available at https://github.com/1HUAN1/RSI-Flow.

A22026-10-15cs.AI

Learning from Interaction Experience via Localized Context Distillation

Zhezheng Hao, Yongkang Yang, Xilin Xia, Jiarui Yu, Yi Liu, Fanjunduo Wei, Peijie Dong, Tianfu Wang, Qiang Lin, Can Wang, Hande Dong, Jiawei…

Learning from interaction experience is essential for enabling large language models to continually improve during deployment. Context distillation offers a promising path: privileged information extracted from experience is provided to a teacher, whose behavior is then distilled into a student model that has no access to it. However, both existing forms of context distillation have limitations. Off-policy context distillation trains the student on teacher-generated trajectories and thus suffers from exposure bias; whereas on-policy context distillation trains on the student's own rollouts, whose states are often uninformative and fail to convey behaviors the student cannot yet produce. To resolve this dilemma, we introduce Localized Context Distillation (LoCD). LoCD first localizes the critical spans of a student trajectory that require revision. The teacher then regenerates only these spans from the student's own prefix, and the student is trained on the regenerated segments. LoCD thereby concentrates learning on the most informative states while keeping exposure bias low. Extensive experiments across math reasoning, code generation, and agent tasks demonstrate that LoCD consistently outperforms strong baselines and achieves more stable training with smaller and sparser parameter updates.

A32026-10-14cs.LG

Hindsight-Privileged RLVR: Single-Rollout Token Credit Without Group Sampling

Wei Chen, Yubing Wu, Zhou Xu, Haoyue Zhang, Delu Zeng, Jiahao Wang, Junmei Yang, Pan tiewen, Jingfeng Liu, Zhiheng Zhou, Bo Li, Min Chen

Reinforcement learning with verifiable rewards (RLVR) trains language models from checked task outcomes, but assigning a terminal reward to individual reasoning tokens or actions remains difficult. Group-relative methods such as Group Relative Policy Optimization (GRPO) sample several attempts per prompt, but lose their reward-based learning signal when all attempts receive the same outcome. We propose Hindsight-Privileged RLVR (HP-RLVR), a single-rollout method that turns a completed trajectory's own future into token-level update weights. The generating policy snapshot scores each eligible token under its causal history and under a hindsight context containing the realized future and verifier feedback. A normalized Jensen--Shannon divergence between the two views weights the reward relative to a historical baseline, yielding bounded token credit for both successful and failed attempts, with no reference solution or learned critic. In paired Qwen3-8B runs over a shared 100-step schedule, HP-RLVR trains 1.21×, 1.46×, and 2.30× faster than GRPO on mathematics, code generation, and ALFWorld, using one eighth the rollouts per prompt. Accuracy is comparable overall, with HP-RLVR ahead on mathematics at all three sizes and GRPO ahead on code and ALFWorld. In ALFWorld, 43.2\% (4B) and 49.0\% (8B) of GRPO's gradient-norm records are zero, versus none for HP-RLVR.

A42026-10-14cs.LG

PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference

Gil Jae Song, Sungjoo Yoo

Long-output inference in large language models suffers from severe KV cache memory and bandwidth bottlenecks. While KV cache quantization alleviates this issue, low-bit quantization (e.g., 2-bit) severely degrades model accuracy. Recent mixed-precision key-cache quantization methods mitigate this degradation by allocating higher precision to important key channels, but estimate channel saliency using only key statistics. Through attention logit perturbation analysis at the RoPE channel-pair level, we show that quantization errors in channel pairs with large query--key pair-norm products contribute significantly to logit variation. Based on this insight, we propose PN-Mix, a mixed-precision key-cache quantization method for RoPE-based decoder-only LLMs, which allocates higher precision to salient channel pairs identified once during the prefill stage and reused throughout decoding. Across standard reasoning, code-generation, and AIME benchmarks, PN-Mix with 2.26-bit average precision achieves comparable or better average accuracy than prior mixed-precision methods using 2.32--2.39 bits. Furthermore, in our Pareto sweep, PN-Mix degrades gracefully down to 1.5 bits, whereas the compared baselines exhibit sharper performance drops below 2.25 bits.

A52026-10-13cs.AI

Beyond ICD Codes: Semantic IDs for Generative Medical Coding

Yihang Cheng, Qian Zhang, Yonghui, Huang, Baoshuo Kan, Junyi Shen

Generative approaches to automatic International Classification of Diseases (ICD) coding use large language models (LLMs) to predict ICD codes from clinical documents. However, original ICD code strings are not ideal generation targets because the ICD hierarchy does not always reflect clinical relationships, and LLMs struggle to interpret and generate these codes. We introduce SemICD, a framework that trains LLMs to generate Semantic IDs (SIDs) instead. SemICD constructs SIDs by hierarchically clustering embeddings of textual code profiles, allowing semantically related codes to share prefixes while each complete SID uniquely identifies one ICD code. To align the SID tokens with clinical semantics, we employ two auxiliary tasks for bidirectional generation between textual code profiles and SIDs. Experiments on MIMIC-III and MIMIC-IV with four LLMs show that SemICD outperforms direct code generation, improving Micro-F1 by 1.34-8.00 points. Our code is available at https://github.com/youwillbethebest/SemICD.

A62026-10-13cs.LG

Verifier-Grounded Agentic RL for Provably Correct Code

George Ma, Yuhao Zhang, Bing He, Zeren Shui, Ferhat Erata, Murali Krishna Ramanathan, Somayeh Sojoudi, Hao Zhou

Reinforcement learning for code usually rewards passing tests, which a model can do while remaining wrong on the inputs the tests do not cover. A program verifier instead decides whether a program satisfies a formal specification for all inputs. Language models are almost always paired with verifiers at inference time only, and turning a verifier into a reward raises two difficulties. It reports only acceptance or rejection, which gives no gradient when every attempt fails, and a policy can cheat it, most simply by telling it to assume what the proof was supposed to establish. We build an agentic environment and reward that address both. The reward reads the verifier's diagnostics rather than its verdict, scoring a candidate by the weighted fraction of proof obligations it discharges, extended with hidden probes from a reference proof's annotations. An integrity audit outranks the verifier, rejecting any candidate that alters the specification or assumes what it was meant to prove. The model works as an agent against a live verifier over many turns, with a separately trained proof specialist as a frozen collaborator. Across Qwen3 models from 1.7B to 14B and a 35B Qwen3.6 mixture-of-experts model, held-out verification on code generation rises from 66.1% to 93.5% on DafnyBench and from 56.3% to 100% on SWE-Proof. This reward leads every other training signal we compare, including a binary pass-or-fail reward trained under otherwise identical settings, with the widest margin on code generation. When the agent must also deliver the repository patch, graded by each project's withheld tests, the fraction of held-out issues it resolves rises from 59.6% to 72.5%. Evaluated unchanged on three verification suites withheld from training, the checkpoints improve on all of them, including one whose language and verifier are absent from training.

A72026-10-13cs.SE

DivAlgo: A systematic evaluation of algorithmic diversity for code generation models

Anubhav Jangra, Smaranda Muresan

Code generation models are actively being used in professional and educational settings. However, the competence of large language models (LLMs) to generate algorithmically diverse code remains largely unknown. We propose the DivAlgo benchmark, a large-scale dataset constituting algorithmically diverse solutions to programming problems. We also introduce a lightweight similarity metric to enable large-scale evaluation. Our experiments across 24 modern code generation models found that uncontrolled model-generated code is significantly less diverse than human solutions (capturing 31% of human diversity), and even the best diverse code generation strategy could achieve only 70% of human diversity. Moreover, while it was possible to steer the models to generate solutions based on human strategies, they were unable to reproduce the diversity of the human solutions. These findings indicate the limitations of code generation models in capturing the algorithmic diversity humans exhibit, and that overexposure to model-generated code can narrow programmers' algorithmic vocabulary.

A82026-10-13cs.LG

Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models

Miao Li, Hanyang Jiang, Hengyu Fu, Baihe Huang, Yating Liu, Yuhang Cai, Changkun Guan, Zhongzhu Chen, Jiecheng Lu, Michael A. Boateng, Pasc…

Diffusion language models (DLMs) enable flexible parallel generation by resolving tokens according to the model's evolving predictability rather than a fixed left-to-right order. This flexibility, however, introduces training choices absent from standard autoregressive teacher forcing: which reference tokens are visible when a target is supervised and how much generation space is allocated. Standard supervised fine-tuning (SFT) typically makes these choices through random masking, independently of inference. Our analysis shows how this training-inference mismatch can misdirect supervision. Random masking can conceal prediction errors or penalize uncertainty that decoding would otherwise resolve, while the attention-visible input length can change which response formulations the model favors, even when the underlying solution is unchanged. Motivated by these findings, we propose Predictability-Aware Canvas Selection and Supervision (PACS), an SFT method. PACS uses the model's own predictability to select each example's attention-visible input length and supervises reference tokens immediately before their confidence-guided revelation. Across reasoning and code-generation benchmarks, PACS consistently achieves the highest mean accuracy among the compared SFT methods, with gains of up to 12.84 percentage points over vanilla SFT on individual benchmarks. Code will be released upon publication.

Foundational papers

  1. 2021
    Usage of large language model for code generation tasks: A review44 citations
  2. 2023
    Planning with large language models for code generation361 citations
  3. 2023
    Codet5+: Open code large language models for code understanding and generation1,166 citations
  4. 2023
    Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation2,721 citations
  5. 2025
    Large language models for code generation: A comprehensive survey of challenges, techniques, evaluation, and applications110 citations
  6. 2026-10
    Latest on arXiv: RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agentcs.AI
Citations per paperCitation counts via the AISA Scholar API
citations: R1 A survey on large lan… 1,589, R2 Planning with large l… 361, R3 Self-planning code ge… 556, R4 Large language models… 110, R5 Usage of large langua… 44, R6 Codet5+: Open code la… 1,166, R7 Is your code generate… 2,721, R8 On the effectiveness … 179, R9 Exploring parameter-e… 180, R10 Large language model-… 109R1 A survey on large lan…1,589R2 Planning with large l…361R3 Self-planning code ge…556R4 Large language models…110R5 Usage of large langua…44R6 Codet5+: Open code la…1,166R7 Is your code generate…2,721R8 On the effectiveness …179R9 Exploring parameter-e…180R10 Large language model-…109
R11,589 citations

A survey on large language models for code generation

J Jiang, F Wang, J Shen, S Kim, S Kim

… Large language models have become transformative tools for automated code generation, … of LLMs for code generation, as illustrated in Figure 1. The landscape of LLMs for …

R22023361 citations

Planning with large language models for code generation

S Zhang, Z Chen, Y Shen, M Ding et al.

… Existing large language model-based code generation pipelines typically use beam search … generate achieve high token-matching-based scores, they often fail to compile or generate …

R3556 citations

Self-planning code generation with large language models

X Jiang, Y Dong, L Wang, Z Fang, Q Shang et al.

… (LLMs) have demonstrated impressive ability in code generation, … planning into code generation to help the model understand … code generation approach with large language models, …

How this page is made. Foundational papers and citation counts come from academic search via the AISA Scholar API, with abstracts read from each paper’s page; recent preprints and monthly volumes come from the public arXiv API (metadata CC0). The digest is written by a language model from those abstracts only — every theme links the papers it draws on, and figures not in the inputs are removed.

Sources & citations

  1. A survey on large language models for code generation 1,589 citations
  2. Planning with large language models for code generation 361 citations
  3. Self-planning code generation with large language models 556 citations
  4. Large language models for code generation: A comprehensive survey of challenges, techniques, evaluation, and applications 110 citations
  5. Usage of large language model for code generation tasks: A review 44 citations
  6. Codet5+: Open code large language models for code understanding and generation 1,166 citations
  7. Is your code generated by chatgpt really correct? rigorous evaluation of large language models for code generation 2,721 citations
  8. On the effectiveness of large language models in domain-specific code generation 179 citations
  9. Exploring parameter-efficient fine-tuning techniques for code generation with large language models 180 citations
  10. Large language model-aware in-context learning for code generation 109 citations
  11. RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent arXiv · 2026-10-15
  12. Learning from Interaction Experience via Localized Context Distillation arXiv · 2026-10-15
  13. Hindsight-Privileged RLVR: Single-Rollout Token Credit Without Group Sampling arXiv · 2026-10-14
  14. PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference arXiv · 2026-10-14
  15. Beyond ICD Codes: Semantic IDs for Generative Medical Coding arXiv · 2026-10-13
  16. Verifier-Grounded Agentic RL for Provably Correct Code arXiv · 2026-10-13
  17. DivAlgo: A systematic evaluation of algorithmic diversity for code generation models arXiv · 2026-10-13
  18. Match the Canvas, Not Just the Answer: Predictability-Aware SFT for Diffusion Language Models arXiv · 2026-10-13

Related