Counterexample to Ziegler's conjecture: explicit simplicial 7-dimensional 0/1-polytope with 14 vertices that is not centrally symmetric, plus enumeration of all such examples in dimension 7.
super hub Mixed citations
GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
Mixed citation behavior. Most common role is background (56%).
114
Pith papers citing it
Background
56% of classified citations
abstract
Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% (e.g., +12% accuracy on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa .
hub tools
citation-role summary
background 18
method 5
baseline 3
other 1
citation-polarity summary
claims ledger
- abstract Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing
authors
co-cited works
representative citing papers
Continual Harness automates online self-improvement for foundation-model embodied agents by refining prompts, sub-agents, skills, and memory within one run, cutting button-press costs on Pokemon Red and Emerald and closing much of the gap to expert harnesses.
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.
ContextLeak is the first empirical framework to audit worst-case information leakage in private in-context learning by inserting identifiable canary tokens and measuring their presence in model outputs.
LLM agents often fail to abstain at the right time in uncertain multi-turn tasks, and the CONVOLVE context engineering method raises timely abstention rates on WebShop from 26.7 to 57.4 without parameter updates.
DeLM decentralizes LLM multi-agent coordination with shared verified context, delivering up to 10.5pp gains on SWE-bench Verified and 5.7pp on LongBench-v2 while cutting cost per task by ~50%.
PACE is a training-free anytime-valid commit gate using testing-by-betting e-processes that controls per-candidate false-commit probability for self-evolving agents and reduces spurious edits compared to greedy acceptance.
EnumGRPO is a self-improving optimizer for agentic query execution that reduces LLM-operator costs by ~317x while improving accuracy by 18% over a hybrid baseline across four databases.
PromptPO shows LLMs can act as black-box policy optimizers for sequential RL when leveraging prior knowledge, matching baselines in exploration and robotics but underperforming in MuJoCo.
CORE distills contrasts between successful and unsuccessful reasoning traces into compact natural-language insights that enable faster model self-improvement on reasoning tasks with fewer rollouts than parametric or other non-parametric baselines.
MOSS performs source-level self-rewriting in agent systems using failure-anchored pipelines and container-based verification, raising OpenClaw mean score from 0.25 to 0.61 in one cycle.
EMO-STA evolves a shared program archive across task families then adapts candidates to targets, outperforming matched-compute single-task evolution in most of eight families while reducing overfitting on low-data tasks like ARC.
Life-Harness evolves reusable interventions from training trajectories to enhance frozen LLM agents on unseen tasks across seven deterministic environments, yielding 88.5% average relative improvement in 116 of 126 model-environment settings.
Residual skill optimization creates complementary Text-to-SQL agents by training each new skill on prior ensemble failures, yielding accuracy gains on Spider2-Lite and transfer to other dialects and tasks.
Evolutionary coding agents achieve most benchmark gains through a small subset of edit types and by cycling previously deleted code lines rather than developing new algorithmic structures.
ReElicit uses LLMs to elicit adaptive feature embeddings for Gaussian process Bayesian optimization of system prompts under aggregate-only feedback, outperforming baselines across ten tasks with a 30-evaluation budget.
PluRule is a new multimodal multilingual benchmark showing that state-of-the-art vision-language models perform only marginally better than a trivial baseline at detecting specific rule violations in pluralistic online communities.
Sleeper memory poisoning enables adversaries to inject fabricated user memories into LLM agents via manipulated external contexts, which are later retrieved and used to induce attacker-intended actions with 60-89% success across models.
AEvo introduces a meta-agent that edits the evolution procedure or agent context based on accumulated state, outperforming baselines by 26% relative improvement on agentic benchmarks and achieving SOTA on open-ended tasks.
Fast-Slow Training uses context optimization as fast weights alongside parameter updates as slow weights to achieve up to 3x better sample efficiency, higher performance, and less catastrophic forgetting than standard RL in continual LLM learning.
AssayBench is a new gene-ranking benchmark for phenotypic CRISPR screens that shows zero-shot generalist LLMs outperform both biology-specific LLMs and trainable baselines on adjusted nDCG.
Evolving-RL jointly optimizes experience extraction and utilization in LLM agents via RL with separate evaluation signals, delivering up to 98.7% relative gains on out-of-distribution tasks in ALFWorld and Mind2Web.
CDS4RAG cyclically optimizes full RAG hyperparameters by distinguishing and alternating between retriever and generator components, boosting performance up to 1.54x over prior methods on benchmarks.
SmellBench is the first benchmark showing LLM agents resolve 47.7% of architectural code smells while accurately spotting false positives, but aggressive repairs often introduce new smells and degrade overall quality.
citing papers explorer
-
A Counterexample to Ziegler's Cross-Polytope Conjecture for Simplicial 0/1-Polytopes
Counterexample to Ziegler's conjecture: explicit simplicial 7-dimensional 0/1-polytope with 14 vertices that is not centrally symmetric, plus enumeration of all such examples in dimension 7.
-
Continual Harness: Online Adaptation for Self-Improving Foundation Agents
Continual Harness automates online self-improvement for foundation-model embodied agents by refining prompts, sub-agents, skills, and memory within one run, cutting button-press costs on Pokemon Red and Emerald and closing much of the gap to expert harnesses.
-
SimWorld Studio: Automatic Environment Generation with Evolving Coding Agent for Embodied Agent Learning
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.
-
ContextLeak: Auditing Leakage in Private In-Context Learning Methods
ContextLeak is the first empirical framework to audit worst-case information leakage in private in-context learning by inserting identifiable canary tokens and measuring their presence in model outputs.
-
Agentic Abstention: Do Agents Know When to Stop Instead of Act?
LLM agents often fail to abstain at the right time in uncertain multi-turn tasks, and the CONVOLVE context engineering method raises timely abstention rates on WebShop from 26.7 to 57.4 without parameter updates.
-
Decentralized Multi-Agent Systems with Shared Context
DeLM decentralizes LLM multi-agent coordination with shared verified context, delivering up to 10.5pp gains on SWE-bench Verified and 5.7pp on LongBench-v2 while cutting cost per task by ~50%.
-
PACE: Anytime-Valid Acceptance Tests for Self-Evolving Agents
PACE is a training-free anytime-valid commit gate using testing-by-betting e-processes that controls per-candidate false-commit probability for self-evolving agents and reduces spurious edits compared to greedy acceptance.
-
Cost-Aware Optimization for Agentic Query Execution
EnumGRPO is a self-improving optimizer for agentic query execution that reduces LLM-operator costs by ~317x while improving accuracy by 18% over a hybrid baseline across four databases.
-
When are LLMs Sufficient Policy Optimizers for Sequential RL Tasks?
PromptPO shows LLMs can act as black-box policy optimizers for sequential RL when leveraging prior knowledge, matching baselines in exploration and robotics but underperforming in MuJoCo.
-
CORE: Contrastive Reflection Enables Rapid Improvements in Reasoning
CORE distills contrasts between successful and unsuccessful reasoning traces into compact natural-language insights that enable faster model self-improvement on reasoning tasks with fewer rollouts than parametric or other non-parametric baselines.
-
MOSS: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems
MOSS performs source-level self-rewriting in agent systems using failure-anchored pipelines and container-based verification, raising OpenClaw mean score from 0.25 to 0.61 in one cycle.
-
Evolutionary Multi-Task Optimization for LLM-Guided Program Discovery
EMO-STA evolves a shared program archive across task families then adapts candidates to targets, outperforming matched-compute single-task evolution in most of eight families while reducing overfitting on low-data tasks like ARC.
-
Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents
Life-Harness evolves reusable interventions from training trajectories to enhance frozen LLM agents on unseen tasks across seven deterministic environments, yielding 88.5% average relative improvement in 116 of 126 model-environment settings.
-
Residual Skill Optimization for Text-to-SQL Ensembles
Residual skill optimization creates complementary Text-to-SQL agents by training each new skill on prior ensemble failures, yielding accuracy gains on Spider2-Lite and transfer to other dialects and tasks.
-
What Do Evolutionary Coding Agents Evolve?
Evolutionary coding agents achieve most benchmark gains through a small subset of edit types and by cycling previously deleted code lines rather than developing new algorithmic structures.
-
Embedding by Elicitation: Dynamic Representations for Bayesian Optimization of System Prompts
ReElicit uses LLMs to elicit adaptive feature embeddings for Gaussian process Bayesian optimization of system prompts under aggregate-only feedback, outperforming baselines across ten tasks with a 30-evaluation budget.
-
PluRule: A Benchmark for Moderating Pluralistic Communities on Social Media
PluRule is a new multimodal multilingual benchmark showing that state-of-the-art vision-language models perform only marginally better than a trivial baseline at detecting specific rule violations in pluralistic online communities.
-
Hidden in Memory: Sleeper Memory Poisoning in LLM Agents
Sleeper memory poisoning enables adversaries to inject fabricated user memories into LLM agents via manipulated external contexts, which are later retrieved and used to induce attacker-intended actions with 60-89% success across models.
-
Harnessing Agentic Evolution
AEvo introduces a meta-agent that edits the evolution procedure or agent context based on accumulated state, outperforming baselines by 26% relative improvement on agentic benchmarks and achieving SOTA on open-ended tasks.
-
Learning, Fast and Slow: Towards LLMs That Adapt Continually
Fast-Slow Training uses context optimization as fast weights alongside parameter updates as slow weights to achieve up to 3x better sample efficiency, higher performance, and less catastrophic forgetting than standard RL in continual LLM learning.
-
AssayBench: An Assay-Level Virtual Cell Benchmark for LLMs and Agents
AssayBench is a new gene-ranking benchmark for phenotypic CRISPR screens that shows zero-shot generalist LLMs outperform both biology-specific LLMs and trainable baselines on adjusted nDCG.
-
Evolving-RL: End-to-End Optimization of Experience-Driven Self-Evolving Capability within Agents
Evolving-RL jointly optimizes experience extraction and utilization in LLM agents via RL with separate evaluation signals, delivering up to 98.7% relative gains on out-of-distribution tasks in ALFWorld and Mind2Web.
-
CDS4RAG: Cyclic Dual-Sequential Hyperparameter Optimization for RAG
CDS4RAG cyclically optimizes full RAG hyperparameters by distinguishing and alternating between retriever and generator components, boosting performance up to 1.54x over prior methods on benchmarks.
-
SmellBench: Evaluating LLM Agents on Architectural Code Smell Repair
SmellBench is the first benchmark showing LLM agents resolve 47.7% of architectural code smells while accurately spotting false positives, but aggressive repairs often introduce new smells and degrade overall quality.
-
Back to the Beginning of Heuristic Design: Bridging Code and Knowledge with LLMs
A knowledge-first approach to LLM-driven automatic heuristic design in combinatorial optimization yields better discovery efficiency, transfer, and generalization than code-centric baselines by formalizing a distortion-compression trade-off.
-
Co-evolving Agent Architectures and Interpretable Reasoning for Automated Optimization
DeInfer reduces parallel inference communication cost for decomposed LLMs by up to 78% by moving collective operations into the low-rank latent space and redesigning KV-cache reconstruction for static graph compatibility.
-
LLM-HYPER: Generative CTR Modeling for Cold-Start Ad Personalization via LLM-Based Hypernetworks
LLM-HYPER treats an LLM as a hypernetwork that outputs feature-wise weights for a linear CTR model from few-shot multimodal ad examples, achieving 55.9% better NDCG@10 than cold-start baselines and successful production deployment.
-
Unlocking Prompt Infilling Capability for Diffusion Language Models
Full-sequence masking in SFT unlocks prompt infilling for masked diffusion language models, producing templates that match or surpass hand-designed ones and transfer across models.
-
Meta-Harness: End-to-End Optimization of Model Harnesses
Meta-Harness discovers improved harness code for LLMs via agentic search over prior execution traces, yielding 7.7-point gains on text classification with 4x fewer tokens and 4.7-point gains on math reasoning across held-out models.
-
When Does Continual Learning Require Learning
Different patterns of environmental change (space vs time) require different LLM update behaviors; no single family of methods—prompts, distillation, RL, or compression—handles all regimes.
-
The Harness Effect: How Orchestration Design Sets the Token Economics of Enterprise Agentic AI
Swapping only the orchestration layer around six AI models cut token cost per task 41% and latency 44% with quality at parity, making orchestration a bigger cost lever than model choice.
-
Teaching LLMs to Recommend and Defer in Underrepresented Epilepsy Care
MANANA prompt-learning framework improves LLM prescription prediction accuracy by 4-8 points on Ugandan epilepsy cohorts and supports selective deferral at 95-99% precision on high-confidence cases.
-
Evolution Fine-Tuning: Learning to Discover Across 371 Optimization Tasks
Evolution Fine-Tuning trains LLMs on 156K trajectories spanning 371 tasks to achieve 10.22% average improvement on 22 held-out optimization tasks and match SOTA on select circle-packing problems when combined with test-time RL.
-
CHIA: An open-source framework for principled, agentic AI-driven hardware/software co-design research
CHIA is a distributed graph-based framework for composing, deploying, and verifying agentic AI hardware/software co-design flows, demonstrated on RTL, simulation, and compiler tasks.
-
Autodata: An agentic data scientist to create high quality synthetic data
An agentic weak–strong Self-Instruct loop, optionally meta-optimized, produces synthetic data that trains small models better than standard CoT Self-Instruct across three domains.
-
Towards Spec Learning: Inference-Time Alignment from Preference Pairs
Proposes compiling preference pairs into readable natural-language specifications for inference-time LLM alignment, claiming outperformance over DPO on dense-preference domains.
-
MAS-PromptBench: When Does Prompt Optimization Improve Multi-Agent LLM Systems?
A new benchmark study finds that prompt optimization can deliver significant gains in multi-agent LLM systems but its effectiveness varies strongly with task, workflow, communication protocol, and team size.
-
MetaPS: Adaptive Programmatic Strategy Selection for Market Agents
MetaPS trains models via simulation rollouts to select from programmatic strategy libraries for market agents, yielding better performance than fixed or direct LLM baselines across model sizes.
-
When Is an LLM Worth It for Hyperparameter Optimization? A Budget-Matched Study on Tabular Data Finds the Warm-Start Is a Default Configuration, Not the Model
On eight PMLB tabular benchmarks, an LLM HPO advisor adds only +0.40 pp CV accuracy beyond a fixed default seed and is overtaken by seeded classical methods within 5-12 evaluations, with no held-out test gain.
-
Building Agent Harnesses for Scientific Curation from Multimodal Sources
An agent harness combining staged task decomposition, multimodal evidence tooling, and artifact-grounded self-improvement scores 81.0 GRAS on multimodal scientific curation, 22.4 points above the strongest baseline — with the caveat that 8 of 23 evaluation papers were used for optimization.
-
SoftSkill: Behavioral Compression for Contextual Adaptation
SoftSkill compresses agent skills into length-32 continuous prefixes via next-token training of soft deltas, yielding 5.2-12.5 point gains over SkillOpt on SearchQA and LiveMath while using far fewer tokens.
-
Recursive Self-Evolving Agents via Held-Out Selection
RSEA adds a strict held-out keep-better gate to recursive self-evolution of agent artifacts, yielding monotone-safe gains or parity with the base ReAct agent on ALFWorld, GAIA, τ-bench, and WebShop.
-
Fantastic Scientific Agents and How to Build Them: AgentBuild for Rietveld Refinement
AgentBuild builds scientific agents from version-controlled contracts for Rietveld refinement, progressing through a signal-to-noise ladder to handle 4-hour scans while exposing workflow limits.
-
MUSE: A Unified Agentic Harness for MLLMs
MUSE is a unified agentic harness that improves off-the-shelf MLLMs on visual spatial planning, perception, multimodal reasoning, and fine-grained discrimination benchmarks through structured execution modules and verifier-guided repair without model retraining.
-
BenchEvolver: Frontier Task Synthesis via Solution-Centric Evolution
BenchEvolver evolves coding problem solutions to generate harder, valid tasks, producing LiveCodeBench-Plus where frontier models score 27.5-62.6% and enabling RL gains on held-out tests.
-
Discovering Cooperative Pipelines: Autoresearch for Sequential Social Dilemmas
Autoresearch agent discovers objective-dependent LLM pipelines for sequential social dilemmas that exceed hand-designed baselines and inject fairness mechanisms only under maximin welfare.
-
Inferring Code Correctness from Specification
TRAILS infers code correctness by aggregating LLM judgments on input-output pairs from category-partitioned specification tests, improving MCC by up to 39% over Zero-Shot COT on LiveCodeBench and CoCoClaNeL.
-
MEMENTO: Leveraging Web as a Learning Signal for Low-Data Domains
MEMENTO framework uses adaptive web exploration via AET and dual-channel memory to acquire domain expertise from interaction trajectories, yielding +25.6% and +36.5% gains over ReAct baselines in sales automation and legal research.
-
Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems
Meta-Team is a collaborative self-evolution framework that turns multi-agent execution experience into reusable improvements at agent, coordination, and team levels, outperforming baselines on six benchmarks.
-
Learnable Assessment Skills for LLM-based Automated Scoring: Rubric Construction via Iterative Optimization
An iterative framework lets LLMs learn procedural assessment skills for rubric construction, improving automated scoring on all ten ASAP-SAS items and often exceeding expert rubrics while showing cross-item transfer.