Chain-of-thought prompting, by including intermediate reasoning steps in few-shot examples, elicits strong reasoning abilities in large language models on arithmetic, commonsense, and symbolic tasks.
mega hub Mixed citations
Evaluating Large Language Models Trained on Code
Mixed citation behavior. Most common role is background (65%).
1306
Pith papers citing it
Background
65% of classified citations
abstract
We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of our problems with 100 samples per problem. Careful investigation of our model reveals its limitations, including difficulty with docstrings describing long chains of operations and with binding operations to variables. Finally, we discuss the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics.
hub tools
citation-role summary
background 161
dataset 64
method 12
other 3
baseline 2
citation-polarity summary
claims ledger
- abstract We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of ou
authors
mega hub controls
export citing contexts JSON
export graph JSON
export full bundle JSON
open full Pith review
annotated reader queued
Recognition alignment
counterfactual ablation
co-cited works
representative citing papers
DataComp-VLM benchmark shows instruction-heavy data mixing outperforms filtering for VLM training, with DCVLM-Baseline achieving 63.6% on 33 tasks for 8B models (+5.4pp over FineVision).
Sumi is an openly released 7B parameter uniform diffusion language model pretrained from scratch on 1.5T tokens that matches autoregressive models on several benchmarks.
PCB-QA is the first QA benchmark for LLMs on printed circuit board designs, with Gemini 3 Flash Preview reaching 93% accuracy on a JSON textual representation.
TheoremBench is a Lean4 benchmark of classical theorems in main and premised forms that evaluates LLM provers on partial progress, coverage, and token efficiency rather than binary success on competition problems.
A new benchmark with cognitive traps shows frontier deep research agents achieve only 13-16% acceptance on expert consulting tasks under combined verifier and rubric criteria.
Mistletoe introduces a stealthy attack on speculative decoding that collapses acceleration by reducing average accepted length while preserving output semantics.
CIDR is a large-scale curated dataset of proprietary industrial source code repositories spanning 138 languages and 373 million lines of code, collected via formal agreements with industry partners.
PDEAgent-Bench is the first multi-metric, multi-library benchmark for AI-generated PDE solvers, evaluating executability, numerical accuracy, and efficiency across DOLFINx, Firedrake, and deal.II.
SimWorld Studio deploys an evolving coding agent to create adaptive 3D environments that co-evolve with embodied learners, delivering 18-point success-rate gains over fixed environments in navigation benchmarks.
PhysicianBench is a new benchmark of 100 physician-reviewed, execution-grounded tasks in live EHR environments where the best LLM agent reaches only 46% success and open-source models reach 19%.
AutoMat benchmark shows current LLM coding agents achieve at most 54.1% success when reproducing computational materials science claims from papers.
MLLMs exhibit a Mirage effect by bypassing circuit diagrams in favor of header semantics for Verilog generation; VeriGround with identifier anonymization and D-ORPO training reaches 46% Functional Pass@1 while refusing blank images at >92%.
StabilizerBench is a new benchmark for evaluating AI agents on generating, optimizing, and making fault-tolerant stabilizer circuits for quantum error correction, with efficient verification and multi-tier scoring.
NLI autonomously discovers a vocabulary of primitive operations and interprets variable-length programs via a neural executor, allowing end-to-end training and gradient-based test-time adaptation that outperforms prior methods on combinatorial generalization tasks.
LLM agents autonomously evolve the ABC logic synthesis tool by iteratively rewriting its source code to achieve better quality-of-results on standard benchmarks while preserving the original interface.
FermiLink is a unified AI agent framework that automates multidomain scientific simulations via separated package knowledge bases and a four-layer progressive disclosure mechanism, reproducing 56% of target figures in benchmarks and generating research-grade results on unpublished problems.
DDIPE poisons LLM agent skills by embedding malicious logic in documentation examples, achieving 11.6-33.5% bypass rates across frameworks while explicit attacks are blocked, with 2.5% evading detection.
Derives an exact unbiased policy gradient for RL post-training of diffusion LLMs via entropy-guided step selection and one-step denoising rewards, achieving state-of-the-art results on coding and logical reasoning benchmarks.
MCP-Atlas is a new benchmark with 1000 tasks on production MCP servers that uses claim-level scoring to evaluate LLM agents on realistic multi-step tool-use competency.
CritPt benchmark shows state-of-the-art LLMs reach only 5.7% average accuracy on full-scale unpublished physics research tasks, rising to about 10% with coding tools.
AIDev is a new open dataset of 456k AI-agent pull requests showing agents submit code faster than humans but with lower acceptance rates and simpler changes.
LLaDA is a scalable diffusion-based language model that matches autoregressive LLMs like LLaMA3 8B on tasks and surpasses GPT-4o on reversal poem completion.
Molmo VLMs trained on newly collected PixMo open datasets achieve state-of-the-art performance among open-weight models and surpass multiple proprietary VLMs including Claude 3.5 Sonnet and Gemini 1.5 Pro.
citing papers explorer
-
CIDR: A Large-Scale Industrial Source Code Dataset for Software Engineering Research
CIDR is a large-scale curated dataset of proprietary industrial source code repositories spanning 138 languages and 373 million lines of code, collected via formal agreements with industry partners.
-
Can Coding Agents Reproduce Findings in Computational Materials Science?
AutoMat benchmark shows current LLM coding agents achieve at most 54.1% success when reproducing computational materials science claims from papers.
-
From Mirage to Grounding: Towards Reliable Multimodal Circuit-to-Verilog Code Generation
MLLMs exhibit a Mirage effect by bypassing circuit diagrams in favor of header semantics for Verilog generation; VeriGround with identifier anonymization and D-ORPO training reaches 46% Functional Pass@1 while refusing blank images at >92%.
-
MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers
MCP-Atlas is a new benchmark with 1000 tasks on production MCP servers that uses claim-level scoring to evaluate LLM agents on realistic multi-step tool-use competency.
-
The Rise of AI Teammates in Software Engineering (SE) 3.0: How Autonomous Coding Agents Are Reshaping Software Engineering
AIDev is a new open dataset of 456k AI-agent pull requests showing agents submit code faster than humans but with lower acceptance rates and simpler changes.
-
RepairAgent: An Autonomous, LLM-Based Agent for Program Repair
RepairAgent autonomously repairs 164 bugs on Defects4J including 39 not fixed by prior techniques by treating an LLM as an agent that invokes tools via a finite state machine and dynamic prompts.
-
Regression Accumulation in Multi-Turn LLM Programming Conversations
Regression accumulation affects 40-73% of 8-turn LLM coding tasks on extended HumanEval+/MBPP+ benchmarks, with verification gates improving final-turn pass rates on prior tests.
-
Decoupling Code Complexity from Newcomer Participation: A Causal Study of AI Coding Agent Adoption in OSS
AI coding agent adoption in OSS projects raises code complexity modestly but produces no causal reduction in newcomer participation per DiD estimates on matched GitHub projects.
-
From Registry to Repository: How AI Agent Skills Are Written, Adapted, and Maintained
Empirical study of 41k+ AI agent skills finds reuse is mostly one-time verbatim copying with 53% never modified afterward and maintenance focused on additive local adaptations.
-
The Illusion of Safety: Multi-Tier Verification of AI vs. Human C++ Code
Multi-tier verification on VULBENCH-CPP shows AI-generated C++ code triggers confirmed runtime violations roughly twice as often as human code, while static analysis misleadingly indicates parity due to code length.
-
Falsification, Not Exposure: An Internally Preregistered Placebo-Controlled Decomposition of Self-Repair Feedback in Frozen Small Code Models
Preregistered placebo-controlled decomposition shows external executable counterexamples drive self-repair gains in small code models more than re-exposure or self-critique.
-
AlgoBench: Benchmarking Algorithmic Adaptation in Code Generation
AlgoBench creates traceable variants of competitive programming problems via constraint shifts that invalidate original algorithms, paired with complexity metrics that reveal LLMs often produce functionally correct but asymptotically unsuitable solutions.
-
An Empirical Study of Security Calibration in Large Language Models for Code
Empirical evaluation of three LLMs finds prevalent overconfidence in insecure code generation, with security calibration outperforming functional calibration but both degrading in repository-level settings.
-
Towards Evaluation of Implicit Software World Models in Coding LLMs
Introduces evaluation of LLMs' implicit software world models via prediction of execution resources on real software tasks, finding modest and brittle performance across models including frontier ones.
-
CodeChat-Eval: Evaluating Large Language Models in Multi-Turn Code Refinement Dialogues
CodeChat-Eval shows LLMs lose 19.2% to 69.2% functional correctness over multi-turn refinement dialogues, with largest drops on logic-level and additive changes.
-
Agentic Generation of AST Transformation Rules for Fixing Breaking Updates
BigBag generates reusable AST transformation rules via LLMs that achieve 78.6% fix rate on 157 breaking dependency updates and 33.3% cross-project transfer overall.
-
Beyond Simpson's Paradox: A Cascade of Confounders in AI Agent Pull-Request Co-Authorship
Stratified analysis of AIDev PRs shows co-authorship effects on AI agent merge rates are artefacts of agent composition, repository selection, and PR commit structure rather than causal benefits.
-
RigorBench: Benchmarking Engineering Process Discipline in Autonomous AI Coding Agents
RigorBench evaluates AI coding agents on process discipline via five pillars and reports 41% higher process scores and 17% better outcome correctness with structured approaches on 30 tasks.
-
The Alignment Problem in Constrained Code Generation
Incomplete constrainers in constrained decoding push LLMs into low-probability program regions, making unconstrained decoding outperform constrained decoding on functional correctness across seven models and three benchmarks.
-
N-Version Programming with Coding Agents
Diverse AI coding agents in N-version programming reduce mean failures from 387.44 to 130.99 in triples on the Launch Interceptor Program, with 11,844 zero-failure units observed across 1M tests.
-
Repository-Level Solidity Code Generation with Large Language Models: From Prompting to Fine-Tuning
Introduces SolidityBench benchmark and SolidityScore metric for repository-level Solidity code generation, finding supervised fine-tuning outperforms prompting, CoT, ICL, and RAG methods on evaluated LLMs.
-
StaminaBench: Stress-Testing Coding Agents over 100 Interaction Turns
StaminaBench evaluates coding agents over 100 procedurally generated change requests to a REST API, finding that tested models fail within 5-6 turns without feedback but improve up to 12x with test feedback and good harnesses.
-
Where Did the Variability Go? From Vibe Coding to Product Lines by Regeneration
Exploratory study of vibe-coded projects shows variability is bound at generation time; proposes VbR as an SPL method using LLMs to generate variant-specific code from specifications.
-
Configuration Smells in AGENTS.md Files: Common Mistakes in Configuring Coding Agents
Presents first catalog of six smells in coding-agent config files, with automated detection heuristics, and reports high prevalence (e.g., Lint Leakage in 62%) from analysis of 100 open-source repos.
-
When LLMs Invent Rust Crates: An Empirical Study of Hallucination Patterns and Mitigation
First empirical study shows crate hallucination in Rust LLMs has consistent rates across models insensitive to parameters and tests prompt-based mitigation.
-
SkelDPO: A Skeleton-Guided Direct Preference Optimization Framework for Efficient Code Generation
SkelDPO improves code generation efficiency by 2-7% over prior DPO methods via joint preference losses on full code and efficiency-critical skeletons.
-
Asuka-Bench: Benchmarking Code Agents on Underspecified User Intent and Multi-Round Refinement
Asuka-Bench is a new benchmark of 50 web tasks with 784 criteria that evaluates 8 LLMs in 2 frameworks on multi-round refinement, finding a 38-point spread in weighted task pass rate and a top score of only 52% after three rounds.
-
ADK Arena: Evaluating Agent Development Kits via LLM-as-a-Developer
ADK Arena evaluates 51 Python ADKs by having an LLM learn each framework's API, write and repair agent code, and run on benchmarks, finding 57% success rate, 5.6x cost variation, no dominant framework, and substitutable information sources.
-
Sakura: An Approach for Generating Complex Tests from Natural Language Test Descriptions
Sakura is a multi-agent system that generates structurally complex tests from NL descriptions, achieving 50-78% higher compilability and 38-66% higher coverage overlap than baselines on 1,464 scenarios from 20 Apache Commons applications.
-
How Generation Architecture Shapes Code Complexity in Multi-Agent LLM Systems: A Paired Study on HumanEval
Six multi-agent architectures for LLM code generation on HumanEval form two complexity clusters separated by 50-130%, with no accuracy advantage for the complex cluster.
-
Knowledge Boundary Probing and Demand-Guided Intervention for LLM-Based Power System Code Generation
PowerCodeBench and a boundary-aware intervention raise LLM accuracy on power-system code generation by 32-56 points across ten open-weight models and four commercial APIs on a 2,000-task benchmark.
-
What Breaks When LLMs Code? Characterizing Operational Safety Failures of Agentic Code Assistants
An empirical study of 547 confirmed safety incidents from GitHub and literature derives a 33-type taxonomy showing constraint violations, destructive actions, and deception dominate in everyday coding-agent use.
-
On the Road to Personalized Code Intelligence: Portraiting and Assisting Developers Based on Their In-IDE Behaviors
VirtualME is a new infrastructure that continuously extracts and interprets in-IDE developer behaviors to build personalized personas, delivering 33.8% better performance on repository-level knowledge Q&A than generic baselines.
-
Efficient and Scalable Provenance Tracking for LLM-Generated Code Snippets
Hybrid vector-search plus fingerprinting pipeline for LLM code provenance achieves Winnowing-level MRR on short snippets and up to 5.4% better on longer ones at logarithmic query time.
-
Towards Demystifying and Repairing LLM-in-the-Loop Vulnerabilities
Authors create LLMCVE dataset of LLM-in-the-loop vulnerabilities and demonstrate that agent-based repair methods achieve low success rates on them, particularly prompt injections at 28.57% Pass@1.
-
Trustworthy Software Project Generation : a Case Study with an Interactive Theorem Prover
An LLM agent with Rocq backend automatically builds a verified RISC-V RV32I interpreter (1859 lines Rocq, 2848 lines extracted C++) that passes 265 tests and 12-hour fuzzing, while a Dafny backend fails.
-
RepoMirage: Probing Repository Context Reasoning in Code Agents with Perturbations
RepoMirage uses semantics-preserving perturbations on SWE-Bench to show code agents lack repository context reasoning, with performance falling sharply on extended structure tasks, and introduces RepoAnchor as a structure-first fix.
-
VISTA: An End-to-End Benchmark for Visual Spec-to-Web-App Coding Agents
VISTA is a new benchmark for end-to-end visual spec-to-web-app generation by LLM agents, featuring five prompt conditions, manual UI annotations, multi-metric evaluation, and results on four agent systems showing partial decoupling of visual and functional performance.
-
SpecBench: Measuring Reward Hacking in Long-Horizon Coding Agents
SpecBench shows frontier coding agents saturate visible test suites but exhibit persistent reward hacking on held-out tests, with the gap growing 28 percentage points per tenfold increase in code size.
-
Code Generation by Differential Test Time Scaling
DiffCodeGen clusters code candidates by behavioral similarity from fuzzing-synthesized inputs and selects the largest cluster's medoid, matching or exceeding prior test-time scaling methods with far less token and time cost.
-
Reversa: A Reverse Documentation Engineering Framework for Converting Legacy Software into Operational Specifications for AI Agents
Reversa is a reverse documentation engineering framework that deploys a multi-agent pipeline to extract implicit rules from legacy software and produce traceable specifications with confidence scores and explicit gaps for human review.
-
SaaSBench: Exploring the Boundaries of Coding Agents in Long-Horizon Enterprise SaaS Engineering
SaaSBench introduces a heterogeneous benchmark for enterprise SaaS engineering and shows that state-of-the-art coding agents fail over 95% of the time before reaching deep business logic due to setup and integration problems.
-
SWE-Chain: Benchmarking Coding Agents on Chained Release-Level Package Upgrades
SWE-Chain provides 155 chained version transitions and 1,660 requirements across 9 Python packages, where frontier agents resolve 44.8% of tasks on average and struggle to preserve functionality across releases.
-
PBT-Bench: Benchmarking AI Agents on Property-Based Testing
PBT-Bench is a new benchmark with 100 property-based testing problems across 40 Python libraries that measures LLM bug recall rates of 42.1-83.4% under guided prompting versus 31.4-76.7% in baseline.
-
AI Harness Engineering: A Runtime Substrate for Foundation-Model Software Agents
The paper defines AI Harness Engineering as a runtime substrate with eleven components and a four-level ladder that reframes agent reliability as a model-harness-environment system property rather than model capability alone.
-
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
-
Constraint Decay: The Fragility of LLM Agents in Backend Code Generation
LLM agents exhibit constraint decay with assertion pass rates dropping substantially as structural requirements increase in multi-file backend code generation across web frameworks.
-
An Empirical Study of Proactive Coding Assistants in Real-World Software Development
Real developer IDE traces differ substantially from LLM simulations in behavior and structure; current proactive assistants are unreliable on real traces, and simulated data cannot substitute for real data in training.
-
HEJ-Robust: A Robustness Benchmark for LLM-Based Automated Program Repair
LLM-based Java program repair models lose over 50% of their bug-fixing success rate when presented with equivalent but syntactically varied buggy code.
-
Social Bias in LLM-Generated Code: Benchmark and Mitigation
LLMs show up to 60.58% social bias in generated code; a new Fairness Monitor Agent cuts bias by 65.1% and raises functional correctness from 75.80% to 83.97%.