AgentFlare

Scholar · code generation large language models

Code Generation with LLMs: Research Digest

10 foundational papers with 364 combined citations, plus the latest arXiv preprints.

Foundational papers10364 citations
Most citedLLMs for science: Usage for code gene…212 citations
arXiv, Sep 202685 papers-31% vs Oct 2025
Newest preprint2026-10-27

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

A12026-10-27cs.AI

Guided-RLVR: Guided Exploration in Reinforcement Learning with Verifiable Rewards

Bo Yang, Zishuai Zhang, Weihua Li, Qinnan Zhang, Sijia Wen, Hainan Zhang

Reinforcement learning with verifiable rewards (RLVR) has shown strong empirical gains in improving the reasoning ability of large language models (LLMs) on tasks such as mathematics and programming. However, recent studies have shown that RLVR rollouts often collapse onto a narrow set of reasoning trajectories, limiting the explored reasoning space. Motivated by findings that thinking tokens at critical steps serve as critical structural elements in LLM reasoning pathways, we propose Guided-RLVR, a novel approach that augments RLVR with a lightweight controller. The controller strategically injects semantically meaningful steering tokens into the sampling distribution, encouraging the model to approach problems from multiple reasoning perspectives and thereby expanding the exploration space. Experiments on mathematical reasoning and code generation benchmarks show that Guided-RLVR consistently outperforms existing RLVR methods while producing more diverse reasoning trajectories.

A22026-10-27cs.LG

HiLo-RD: Cross-Budget Self-Distillation for Masked Diffusion Language Models

Xiang Li, Chengting Yu, Jungang Li, Yilun Zheng, Junzhuo Li, Xingyao Zhang, Wenbo Su, Xuming Hu

We propose HiLo-RD (High-to-Low Budget Representation Distillation) to improve masked diffusion language models (MDLMs) under a limited generation budget. Unlike autoregressive models, MDLMs choose both what to generate and where to commit it. This creates two coupled challenges: predicting useful content before reasoning is complete, and deciding when to fix answer tokens that shape later predictions. We identify budget-induced premature commitment: even when a correct solution fits, the model can fix answer fragments near the window boundary before resolving intermediate steps. These fragments become context for later predictions and redirect subsequent reasoning. Existing post-training methods do not directly transfer the stronger trajectories that the same model follows with a larger window. HiLo-RD addresses this gap by (1) pairing large-window teacher states with cropped small-window views and (2) matching their hidden representations at committed positions. This supervision guides both token prediction and commitment order without an external teacher. Across mathematics benchmarks, HiLo-RD improves low-budget accuracy and shortens responses; these gains extend to code generation without code-domain training. On GSM8K, accuracy rises by 9.4 percentage points as mean readable length falls by about 25%. Attention diagnostics and fixed-budget order interventions support the proposed mechanism. Together, these results show that large-window reasoning can improve small-window decoding.

A32026-10-27cs.CL

CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models

Jingpu Yang, Tianhao Li, Fengxian Ji, Yufeng Wang, Shuxin Zheng, Xiaoqing Zhang

Continual learning requires large language models to acquire new knowledge and capabilities while retaining what they have learned. Fine-tuning and replay support acquisition and retention, but deciding whether incoming examples need supervision and which historical examples to revisit remains a challenge. We propose Compatibility-Aware Memory Replay (CAMR), an adaptive framework that jointly controls current-example supervision and historical replay. CAMR combines compatibility checks with a learned memory gate to guide an online policy over three actions: learning incoming examples with general replay, jointly reinforcing compatible related examples, and skipping direct supervision of verified mastered duplicates. The policy learns from acquisition, retention, and training-cost feedback under a bounded per-arrival replay budget. We evaluate CAMR with Qwen3-8B-Base on continual factual updating and cross-task and cross-domain learning. On WikiBigEdit, CAMR consistently leads evaluated methods in Original Match from 3K to 100K facts, reaching 93.66% after just 3K facts and surpassing Full-history Replay (LoRA) after 100K facts. It also achieves the highest final scores on all four selected TRACE tasks and maintains or improves earlier-task performance relative to post-acquisition levels in a sequence spanning financial question answering, mathematical reasoning, code generation, and summarization.

A42026-10-27cs.SE

Benchmarking LLM Agents for Behaviorally Adequate Test Generation from Software Requirements

Binhang Qi, Yiming Liu, Yun Lin, Yangjie Zhou, Hailong Sun, Jin Song Dong

As large language model agents increasingly compile natural-language requirements into repository-level code, requirement-to-test generation grows in importance to provide validation constraints (or harness) to guide code generation. Yet it remains unclear whether current agents can generate tests that validate the required behavior when the implementation of that behavior is withheld. Existing test-generation benchmarks mostly focus on white-box testing, whereas the requirement--test-related study usually requires evaluation on semantic alignment between the given documents and the generated tests. We introduce Req2Test-Bench, a benchmark containing 5,086 execution-validated instances recovered from developer-authored behavior-driven development (BDD) artifacts in 8 open-source projects, spanning 5 programming languages and 4 BDD frameworks. Each instance pairs a natural-language requirement in Gherkin with its developer-written Test Implementation. Given the requirement, target interfaces of the required behavior, an agent must generate a complete, potentially multi-file Test Implementation. Evaluation of 8 agent--model--reasoning configurations reveals a substantial gap between execution success and behavioral adequacy, which is measured strictly as kill-completeness: a kill-complete test kills every mutant that the ground-truth test kills. The best-performing configuration, Codex with DeepSeek-V4.1-Flash, produces passing tests for 82.8% of instances, yet only 33.8% are behaviorally adequate. We find that (1) scaling up agents yields sharply diminishing returns in behavioral adequacy relative to generation cost; (2) under-specified requirements cause the information gap, including ambiguous setup and implicit expected outcomes; and (3) requirement-grounded harnesses that steer agent non-determinism offer a promising direction for improvement.

A52026-10-26cs.CL

Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models

Aidar Myrzakhan, Zhiqiang Shen

Diffusion large language models (dLLMs) can generate multiple tokens in parallel, but often require many decoding rounds to complete a response. Efficient generation depends not only on which tokens are easy to predict, but also on how committing to them helps predict the remaining positions. A highly predictable token may provide little useful context, while revealing a less predictable token can help several other positions become predictable sooner. We propose dSpark, a new schedule distillation-based framework for learning when to commit tokens for faster diffusion language generation. To achieve this, we construct unmasking schedules that prioritize tokens whose reveal makes other positions easier to predict, reducing uncertainty earlier in decoding. We then distill these schedules into the model to enable faster parallel sampling, with a KL-divergence regularizer to help preserve generation quality. We extensively evaluate our approach across mathematical reasoning and code generation tasks. On GSM8K, our method achieves remarkable 10.12× and 5.12× speedups over the various baselines on LLaDA-8B-Instruct and Dream-7B-Instruct, respectively, while maintaining superior task performance. Our code is available at https://github.com/VILA-Lab/dSpark.

A62026-10-25cs.CL

Transformer to GPT-5: Advancements, Limitations, and Future Directions in Large Language Model Design

Harsidak Singh Banwait, Chander Prabha

Large Language Models (LLMs) have revolutionized Artificial Intelligence (AI) by their extraordinary capabilities, like contextual understanding, few-shot/zero-shot learning, code generation, ability to reason and problem-solving. Recent studies have shown immense work on LLMs. Yet, there is limited work on the thorough taxonomy of LLMs. This study would showcase a detailed classification of LLMs, starting with the earlier models like BERT and leading to LLMs like GPT, Gemini, Claude and Grok. We would delve into topics like datasets used to train LLMs, computational power used during training, their hardware specifications, architecture and parameters. Further, we would also look at the limitations, computational challenges and open problems as well. This study would primarily focus on the timeline of LLMs, signifying their growth over the period. Findings reveal a clear pattern: models that optimize the ratio between architectural complexity and data quality exhibit stronger reasoning stability and better alignment, without proportionally increasing the computational load. This desired balance between scaling and sustainability is the defining characteristic of the current generation of AI.

A72026-10-24cs.CR

Secure and Functional Code Generation with Self-Evaluation

Md Afif Al Mamun, Gias Uddin

Large Language Models (LLMs) can generate functionally correct code that remains insecure. Security-focused fine-tuning can be expensive and is not always available, motivating inference-time approaches. Security-oriented prompting and retrieval can reduce vulnerabilities, but may also remove required behavior or apply defenses incompatible with the task. The challenge is to identify relevant security knowledge and use it to improve security without sacrificing functionality. We present SecGent, an inference-time framework that retrieves weakness context from the Common Weakness Enumeration (CWE) database and applies it to secure code generation. By connecting task specifications to relevant textual CWE records, SecGent provides task-specific security guidance without relying on retrieved vulnerable or patched code examples. Complementary functional and security tests check the resulting implementation, and their execution feedback guides repairs toward satisfying both objectives. We evaluate SecGent with four open-source LLMs and Claude Sonnet 5.5 across five programming languages on the CWEval benchmark. It achieves the highest joint functional--security success against four inference-time baselines for all five models. Across the open-weight models, mean functional-security@1 (FS@1) improves by 15.5% over RAG, with gains in both functionality and security, and by 19.8% over the strongest prompt optimizer compiled for each model. On Sonnet, FS@1 reaches 92.1%, outperforming the strongest baseline AutoGen by 4.8%. In a separate comparison, SecGent outperforms Claude Code with security-focused prompting by over 6% in FS@1. Feedback-guided repair further improves FS@1 by up to 6.7% over generation without feedback, primarily by recovering functionality while checking security.

A82026-10-23cs.LG

Accordion: Elastic-Depth Inference via Fixed-Prefix Layer Looping

Yudong Zhang, Wei Ding, Kaili Zheng, Yuchen Ding, Zhenyu Zhang

Large language models serve applications with diverse inference budgets, but training and maintaining separate models for each budget incurs substantial costs. Recent Looped Transformers enable flexible inference computation through parameter reuse, but maintaining strong performance across different inference budgets remains an open challenge. We introduce Accordion, a mixture-of-experts-based model that supports five inference budgets through fixed-prefix layer looping. Each budget follows a pre-defined execution path, while the minimum budget retains the unlooped model. Joint pretraining samples one budget per micro-batch and optimizes its next-token prediction loss without additional cross-budget consistency objectives. We further preserve these inference budgets through supervised fine-tuning and reinforcement learning, where parameter-group merging and budget-matched multi-teacher on-policy distillation integrate three domain experts into one model. After 3.6T tokens of pretraining, increasing inference depth improves Accordion-6B-A0.5B's five-category macro-average across 37 tasks by 3.01 percentage points, with the middle budget capturing 89% of this gain using half as many additional layer executions. The final model surpasses each domain expert on its corresponding task-and-budget average, while achieving strong instruction-following performance across all budgets and retaining depth-dependent gains in mathematical reasoning and code generation.

Foundational papers

  1. 2024
    Imperfect code generation: Uncovering weaknesses in automatic code generation by large language models10 citations
  2. 2026-10
    Latest on arXiv: Guided-RLVR: Guided Exploration in Reinforcement Learning with Verifiable Rewardscs.AI
Citations per paperCitation counts via the AISA Scholar API
citations: R1 Aligning Requirement … 23, R2 Fine-tuning mistral 7… 48, R3 A framework for train… 12, R4 Where are large langu… 29, R5 Revisiting chain-of-t… 17, R6 What Makes Effective … 0, R7 Imperfect code genera… 10, R8 Evaluating the impact… 4, R9 Quantizing large lang… 9, R10 LLMs for science: Usa… 212R1 Aligning Requirement …23R2 Fine-tuning mistral 7…48R3 A framework for train…12R4 Where are large langu…29R5 Revisiting chain-of-t…17R6 What Makes Effective …0R7 Imperfect code genera…10R8 Evaluating the impact…4R9 Quantizing large lang…9R10 LLMs for science: Usa…212
R10212 citations

LLMs for science: Usage for code generation and data analysis

M Nejjar, L Zacharias, F Stiehle et al.

… Bias assessment and mitigation in LLM-based code generation. arXiv preprint arXiv:… experience: evaluating the usability of code generation tools powered by large language models. In: …

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. Aligning Requirement for Large Language Model's Code Generation 23 citations
  2. Fine-tuning mistral 7b large language model for python query response and code generation: A parameter efficient approach 48 citations
  3. A framework for training large language models for code generation via proximal policy optimization 12 citations
  4. Where are large language models for code generation on github? 29 citations
  5. Revisiting chain-of-thought in code generation: Do language models need to learn reasoning before coding? 17 citations
  6. What Makes Effective Knowledge Distillation for Large Language Models in Code Generation? 0 citations
  7. Imperfect code generation: Uncovering weaknesses in automatic code generation by large language models 10 citations
  8. Evaluating the impact of post-training quantization on large language models for code generation 4 citations
  9. Quantizing large language models for code generation: A differentiated replication 9 citations
  10. LLMs for science: Usage for code generation and data analysis 212 citations
  11. Guided-RLVR: Guided Exploration in Reinforcement Learning with Verifiable Rewards arXiv · 2026-10-27
  12. HiLo-RD: Cross-Budget Self-Distillation for Masked Diffusion Language Models arXiv · 2026-10-27
  13. CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models arXiv · 2026-10-27
  14. Benchmarking LLM Agents for Behaviorally Adequate Test Generation from Software Requirements arXiv · 2026-10-27
  15. Learning When to Commit: Schedule Distillation for Fast Diffusion Language Models arXiv · 2026-10-26
  16. Transformer to GPT-5: Advancements, Limitations, and Future Directions in Large Language Model Design arXiv · 2026-10-25
  17. Secure and Functional Code Generation with Self-Evaluation arXiv · 2026-10-24
  18. Accordion: Elastic-Depth Inference via Fixed-Prefix Layer Looping arXiv · 2026-10-23

Related