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.
super hub Mixed citations
Program Synthesis with Large Language Models
Mixed citation behavior. Most common role is background (52%).
534
Pith papers citing it
Background
52% of classified citations
abstract
This paper explores the limits of the current generation of large language models for program synthesis in general purpose programming languages. We evaluate a collection of such models (with between 244M and 137B parameters) on two new benchmarks, MBPP and MathQA-Python, in both the few-shot and fine-tuning regimes. Our benchmarks are designed to measure the ability of these models to synthesize short Python programs from natural language descriptions. The Mostly Basic Programming Problems (MBPP) dataset contains 974 programming tasks, designed to be solvable by entry-level programmers. The MathQA-Python dataset, a Python version of the MathQA benchmark, contains 23914 problems that evaluate the ability of the models to synthesize code from more complex text. On both datasets, we find that synthesis performance scales log-linearly with model size. Our largest models, even without finetuning on a code dataset, can synthesize solutions to 59.6 percent of the problems from MBPP using few-shot learning with a well-designed prompt. Fine-tuning on a held-out portion of the dataset improves performance by about 10 percentage points across most model sizes. On the MathQA-Python dataset, the largest fine-tuned model achieves 83.8 percent accuracy. Going further, we study the model's ability to engage in dialog about code, incorporating human feedback to improve its solutions. We find that natural language feedback from a human halves the error rate compared to the model's initial prediction. Additionally, we conduct an error analysis to shed light on where these models fall short and what types of programs are most difficult to generate. Finally, we explore the semantic grounding of these models by fine-tuning them to predict the results of program execution. We find that even our best models are generally unable to predict the output of a program given a specific input.
hub tools
citation-role summary
background 58
dataset 41
method 4
other 2
citation-polarity summary
claims ledger
- abstract This paper explores the limits of the current generation of large language models for program synthesis in general purpose programming languages. We evaluate a collection of such models (with between 244M and 137B parameters) on two new benchmarks, MBPP and MathQA-Python, in both the few-shot and fine-tuning regimes. Our benchmarks are designed to measure the ability of these models to synthesize short Python programs from natural language descriptions. The Mostly Basic Programming Problems (MBPP) dataset contains 974 programming tasks, designed to be solvable by entry-level programmers. The M
authors
co-cited works
representative citing papers
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.
BoLT is a benchmark of surrogate models fitted to real LLM experiment data that enables evaluation of Bayesian and black-box optimization methods on multi-fidelity, multi-objective, high-dimensional LLM tasks.
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.
Backdoored model code enables deterministic, verifiable stealing of sparse secrets during local LLM fine-tuning via tensor-rule matching and gradient injection, achieving over 98% strict attack success rate while bypassing DP-SGD and auditing defenses.
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.
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.
Molmo2 delivers state-of-the-art open-weight video VLMs with new grounding datasets and training methods that outperform prior open models and match or exceed some proprietary ones on pointing and tracking tasks.
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.
ExCyTIn-Bench is the first benchmark of 7542 questions from Microsoft Sentinel threat investigation graphs, where the best LLM agent achieves a reward of 0.606.
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.
Language models generate robot policy code from natural language commands via few-shot prompting, enabling spatial-geometric reasoning, generalization, and precise control on real robots.
Training language models to generate intermediate computation steps on a scratchpad enables them to perform multi-step tasks such as long addition and arbitrary program execution that they otherwise fail at.
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.
Multi-agent LLMs generate and verify 14,073 deterministic reaction rules from 665,901 patents, enabling 97.7% classification of unseen reactions with finer resolution than fixed proprietary systems.
ELDR reduces median TPOT by 5.9-13.9% in PD-disaggregated MoE serving via expert signatures from prefill, K-means partitioning, and locality-band routing with KV-co-indexed signature cache.
FRAME adds a learnable fractional-Fourier order per expert in a MoE-LoRA setup so that low-rank updates are placed in the domain where they are most compact, yielding gains over fixed-domain baselines on LLaMA-3.1-8B and Qwen2.5-7B.
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.
AxDafny achieves 92.7% verification success on DafnyBench (6.5 points above prior proof-hint baselines) via verifier-guided repair and introduces the LCB-Pro-Dafny benchmark of 250 problems.
Preregistered placebo-controlled decomposition shows external executable counterexamples drive self-repair gains in small code models more than re-exposure or self-critique.
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.
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.
citing papers explorer
-
TEAM: Temporal-Spatial Consistency Guided Expert Activation for MoE Diffusion Language Model Acceleration
TEAM accelerates MoE dLLMs up to 2.2x by exploiting temporal-spatial consistency in expert routing to accept more tokens with fewer activations.
-
Software Self-Extension with SelfEvolve: an Agentic Architecture for Runtime Code Generation
SelfEvolve achieves 92.7% Pass@1 success on 11 runtime self-extension tasks and outperforms baselines like AutoGen by 61.8% with statistical significance.
-
RubberDuckBench: A Benchmark for AI Coding Assistants
RubberDuckBench shows top AI models score around 68% on real GitHub coding questions, rarely answer completely correctly, and hallucinate in 58% of responses on average.
-
EVM-QuestBench: An Execution-Grounded Benchmark for Natural-Language Transaction Code Generation
EVM-QuestBench is a new execution-grounded benchmark with 107 tasks that dynamically evaluates LLMs on generating safe EVM transaction scripts from natural language, revealing large gaps between atomic and composite task performance.
-
SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
-
MURPHY: Feedback-Aware GRPO with Retrospective Credit Assignment for Multi-Turn Code Generation
MURPHY improves code generation pass rates by up to 6% through retrospective credit assignment on multi-turn feedback trees using max or mean reward propagation.
-
Do AI Models Dream of Faster Code? An Empirical Study on LLM-Proposed Performance Improvements in Real-World Software
LLMs propose volatile performance improvements on real-world Java tasks that lag human developers on average, showing algorithmic benchmarks overestimate capabilities.
-
ContractEval: A Benchmark for Evaluating Contract-Satisfying Assertions in Code Generation
ContractEval benchmark on 364 tasks shows code LLMs achieve 75-82% functional pass@1 but 0% contract satisfaction under standard prompting, rising only to 23-41% with explicit contracts.
-
SciML Agents: Write the Solver, Not the Solution
LLMs prompted with domain knowledge can generate runnable, numerically valid code for stiff and non-stiff ODEs on new diagnostic and 1000-task benchmarks.
-
Guidelines for Empirical Studies in Software Engineering involving Large Language Models
The paper delivers a taxonomy of seven LLM study types in software engineering along with eight guidelines that separate mandatory requirements from recommended practices to address reproducibility challenges.
-
Data Mixing Agent: Learning to Re-weight Domains for Continual Pre-training
An RL agent learns domain re-weighting policies from evaluation feedback to improve balanced performance in continual pre-training of LLMs across source and target domains.
-
CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward
CAD-Coder generates valid CadQuery scripts from text via supervised fine-tuning followed by reinforcement learning with geometric Chamfer Distance rewards and chain-of-thought planning.
-
PRIMETIME : Limits of LLMs in Temporal Primitives
PRIMETIME generator reveals that LLM datetime parsing and arithmetic primitives are individually unreliable but fully learnable via fine-tuning, enabling frontier-level accuracy on event planning with small LoRA models.
-
Multi-SWE-bench: A Multilingual Benchmark for Issue Resolving
Multi-SWE-bench provides 1,632 high-quality issue-resolving instances across Java, TypeScript, JavaScript, Go, Rust, C, and C++ for evaluating LLMs on codebase modifications.
-
SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
SWE-RL uses RL on software evolution data to train LLMs achieving 41% on SWE-bench Verified with generalization to other reasoning tasks.
-
Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
A recurrent-depth architecture enables language models to improve reasoning performance by iterating computation in latent space, achieving gains equivalent to much larger models on benchmarks.
-
Semantic Integrity Matters: Benchmarking and Preserving High-Density Reasoning in KV Cache Compression
KV cache compression causes task-dependent degradation in high-density reasoning due to disrupted CoT links; ShotKV mitigates this by preserving few-shot examples as indivisible semantic units through phase separation, delivering 9-18% accuracy gains and 11% latency reduction.
-
DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads
DuoAttention identifies retrieval heads requiring full KV cache and streaming heads using constant-length cache to reduce memory and latency in long-context LLM inference.
-
MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering
MLE-bench evaluates frontier language models as ML engineering agents on 75 Kaggle competitions, with the top setup (o1-preview + AIDE) reaching bronze medal level in 16.9% of tasks.
-
Towards Agentic Runtime Healing
Healer uses LLMs to dynamically generate and execute runtime error-handling code, with GPT-4 recovering from 72.8% of errors across four datasets.
-
DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
DeepSeek-V2 delivers top-tier open-source LLM performance using only 21B active parameters by compressing the KV cache 93.3% and cutting training costs 42.5% via MLA and DeepSeekMoE.
-
AgentCoder: Multi-Agent-based Code Generation with Iterative Testing and Optimisation
A three-agent loop of code generation, test creation, and execution feedback lifts pass@1 to 96.3% on HumanEval and 91.8% on MBPP for GPT-4 while using roughly half the tokens of prior state-of-the-art.
-
GAIA: a benchmark for General AI Assistants
GAIA benchmark shows humans at 92% accuracy on simple real-world questions far outperform current AI systems at 15%, proposing this gap as a key milestone for general AI.
-
RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems
RepoBench is a new benchmark with retrieval, completion, and pipeline tasks to evaluate code auto-completion systems on entire repositories instead of single files.
-
Reflexion: Language Agents with Verbal Reinforcement Learning
Reflexion lets LLM agents improve via stored verbal reflections on task feedback, reaching 91% pass@1 on HumanEval and outperforming prior GPT-4 results.
-
Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks
PoT prompting improves numerical reasoning by having language models write programs executed by a computer instead of performing calculations in natural language chains of thought, with an average 12% gain over CoT.
-
CodeT: Code Generation with Generated Tests
CodeT improves code generation accuracy by using the same model to create test cases and then selecting solutions via output agreement on those tests, raising HumanEval pass@1 from 47% to 65.8%.
-
InCoder: A Generative Model for Code Infilling and Synthesis
InCoder is the first generative model to directly perform zero-shot code infilling via bidirectional context from a masked-then-appended training scheme, matching left-to-right models on synthesis while improving on type inference, comment generation, and variable renaming.
-
MolViBench: Evaluating LLMs on Molecular Vibe Coding
MolViBench is the first benchmark designed to evaluate LLMs on generating executable programs for molecular tasks in drug discovery.
-
OpenClassGen: A Large-Scale Corpus of Real-World Python Classes for LLM Research
OpenClassGen supplies 324,843 real-world Python classes with self-contained skeletons and static metrics to support LLM class generation research and evaluation.
-
kNNGuard: Turning LLM Hidden Activations into a Training-Free Configurable Guardrail
kNNGuard classifies prompts using multi-layer kNN on LLM hidden activations from 50 examples, matching or exceeding fine-tuned guardrails in F1 while running 2.7x to 10x faster with no training required.
-
Prompt Coverage Adequacy
Prompt Coverage Adequacy, measured via attention boosting in LLMs, is associated with fault detection and uncovers over 30% more faults than traditional code coverage when guiding test generation across two datasets.
-
PACE: A Proxy for Agentic Capability Evaluation
PACE builds proxy benchmarks from non-agentic instances via relevance and global selection plus regression to predict agentic scores with MAE under 4%, Spearman correlation above 0.80, and 85% ranking accuracy at under 1% cost.
-
Underspecification does not imply Incoherence: The Risks of Semantic Collapse in Coding Models
Coding LLMs exhibit detrimental semantic collapse on underspecified prompts by producing consistent but incorrect code rather than incoherent variations, affecting 3-32% of tasks across MBPP, HumanEval, and LiveCodeBench.
-
EPnG: Adaptive Expert Prune-and-Grow for Parameter-Efficient MoE Fine-tuning
EPnG reallocates LoRA capacity in MoE models by pruning experts with low router gate probabilities and expanding high-importance ones via rank growth, outperforming standard LoRA and nearing full fine-tuning performance with 0.55-0.72% parameters updated.
-
OPINE-World: Programmatic World Modeling with Ontology-error-Prioritized Interactive Exploration
OPINE-World learns programmatic world models from interaction using dual LLM agents and ontology-error exploration, solving 20 of 25 ARC-AGI-3 games without per-game training.
-
Benchmarking Code Improvement with Progressive, Adaptive, and Interactive Feedback
PAIR-Bench defines a progressive hinting protocol with failure-region and hint-depth controls to measure LLM code refinement trajectories in detail.
-
Are Performance-Optimization Benchmarks Reliably Measuring Coding Agents?
Audit of GSO, SWE-Perf and SWE-fficiency reveals that reference patches satisfy validity rules across machines for only 39/102, 11/140 and 411/498 tasks respectively, public submissions beat references on 85.3% of replay-valid tasks, and scoring rules cause ranking disagreements.
-
ClarifyCodeBench: Evaluating LLMs on Clarifying Ambiguous Requirements for Code Generation
ClarifyCodeBench is a new benchmark with manual annotations and two metrics showing that LLMs strong at code generation are weak at clarifying ambiguous requirements, with performance worsening as ambiguity density rises.
-
SLIM-RL: Risk-Budgeted Random-Masking RL for Diffusion LLMs Without Trajectory Slicing
SLIM-RL matches or exceeds TraceRL performance on MATH500, GSM8K, MBPP and HumanEval for diffusion LLMs by risk-budgeted random-masking RL without trajectory slicing.
-
Benchmarking Large Language Models on Floating-Point Error Classification
Introduces InterFLOPBench benchmark and evaluates 14 LLMs on multi-label classification of six floating-point error categories in C code, with top models exceeding 0.88 overall F1 but lower scores on subtle errors like underflow.
-
SWE-Router: Routing in Multi-turn Agentic Software Engineering Tasks
SWE-Router introduces trajectory-conditioned value-based routing for LLM agents on SWE tasks, with a Bayes-optimality theorem and empirical cost savings while retaining most strong-model performance.
-
SWE-Together: Evaluating Coding Agents in Interactive User Sessions
Introduces SWE-Together benchmark from 109 real repository tasks, using an LLM user simulator to evaluate coding agents on success rate and corrective turns needed.
-
SrDetection: A Self-Referential Framework for Data Leakage Detection in Code Large Language Models
SrDetection detects data leakage in Code LLMs via contrast between original benchmark samples and their semantic variants, reporting F1 gains of 21.52 (gray-box) and 14.46 (black-box) over baselines in a controlled testbed.
-
Breaking the Rounding Trap: Securing LLMs against Quantization-Conditioned Backdoors
QuantGuard is a pre-quantization method using differentiable rounding controls, error-guided reversal constraints, output consistency, and weight regularization on a small calibration set to suppress quantization-conditioned backdoors while preserving performance.
-
Multi-Block Diffusion Language Models
MBD-LMs raise average tokens per forward pass from 3.47 to 6.19 (and to 9.34 with DMax) via multi-block teacher forcing and optimized parallel decoding while holding or slightly improving accuracy on math and code tasks.
-
Data and Evaluation Closed-Loop for Model Capability Enhancement
Proposes capability slices with dual taxonomies and mapping rules to form a closed loop converting benchmark failures into targeted data interventions, validated via two opposing case studies on BBH and math reasoning.
-
When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs
Experiments across code LLMs show no-review collapses fastest, human-gated filters slow collapse, and AI self-gates lose effect over time, degenerating to ungated self-training under self-confirming acceptance as proven via gated distributional reweighting and spectral analysis.
-
Agent-as-a-Router: Agentic Model Routing for Coding Tasks
Agent-as-a-Router turns static LLM routing into an iterative C-A-F loop that accumulates execution feedback to lower cumulative regret on coding tasks.
-
Unlocking LLM Code Correction with Iterative Feedback Loops
Empirical evaluation finds reasoning LLMs improve code correction across iterations using execution feedback and outperform non-reasoning models, with syntactic and runtime errors easier to fix than logical ones.