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.
hub Mixed citations
TextGrad: Automatic "Differentiation" via Text
Mixed citation behavior. Most common role is background (50%).
33
Pith papers citing it
Background
50% of classified citations
abstract
AI is undergoing a paradigm shift, with breakthroughs achieved by systems orchestrating multiple large language models (LLMs) and other complex components. As a result, developing principled and automated optimization methods for compound AI systems is one of the most important new challenges. Neural networks faced a similar challenge in its early days until backpropagation and automatic differentiation transformed the field by making optimization turn-key. Inspired by this, we introduce TextGrad, a powerful framework performing automatic ``differentiation'' via text. TextGrad backpropagates textual feedback provided by LLMs to improve individual components of a compound AI system. In our framework, LLMs provide rich, general, natural language suggestions to optimize variables in computation graphs, ranging from code snippets to molecular structures. TextGrad follows PyTorch's syntax and abstraction and is flexible and easy-to-use. It works out-of-the-box for a variety of tasks, where the users only provide the objective function without tuning components or prompts of the framework. We showcase TextGrad's effectiveness and generality across a diverse range of applications, from question answering and molecule optimization to radiotherapy treatment planning. Without modifying the framework, TextGrad improves the zero-shot accuracy of GPT-4o in Google-Proof Question Answering from $51\%$ to $55\%$, yields $20\%$ relative performance gain in optimizing LeetCode-Hard coding problem solutions, improves prompts for reasoning, designs new druglike small molecules with desirable in silico binding, and designs radiation oncology treatment plans with high specificity. TextGrad lays a foundation to accelerate the development of the next-generation of AI systems.
hub tools
citation-role summary
background 3
method 2
baseline 1
citation-polarity summary
representative citing papers
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.
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
Full factorial testing of five LLM agent components reveals that the complete 'All-In' combination is consistently outperformed by smaller subsets due to cross-component interference, with optimal subsets being task- and scale-dependent.
TSCG compiles JSON tool schemas into token-efficient structured text, raising tool-use accuracy for small LLMs from 0% to 84.4% on benchmarks while cutting tokens by 52-57%.
RosettaSearch applies LLM-driven multi-objective search at inference time to improve backbone-conditioned protein sequences, recovering designs with 18-68% better structural fidelity and 2.5x higher success rates than single-pass models like LigandMPNN.
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.
FVRuleLearner introduces an Operator Reasoning Tree to learn operator-specific rules that improve natural-language to SystemVerilog assertion generation, raising syntax correctness by 3.95% and functional correctness by 31.17% over baselines.
DeepVerifier enables self-evolving deep research agents via rubric-guided verification at test time, delivering 8-11% accuracy gains on GAIA and XBench-DeepSearch subsets.
Meta Agent Search uses a meta-agent to iteratively program novel agentic systems in code, producing agents that outperform state-of-the-art hand-designed ones across coding, science, and math while transferring across domains and models.
PIVOT refines LLM agent trajectories through plan-inspect-evolve-verify stages using environment feedback, yielding up to 94% relative gains in constraint satisfaction and 3-5x token efficiency over prior refinement methods.
MT-JailBench is a modular benchmark that standardizes evaluation of multi-turn jailbreaks to identify key success drivers and enable stronger combined attacks.
AgentSlimming compresses graph-structured multi-agent systems by estimating agent importance and removing or replacing low-value agents, cutting token costs by up to 78.9% with negligible performance loss.
FlashEvolve accelerates LLM agent self-evolution via asynchronous stage orchestration and inspectable language-space staleness handling, reporting 3.5-4.9x proposal throughput gains over synchronous baselines on GEPA workloads.
TPGO represents multi-agent systems as graphs of textual parameters and applies group relative optimization to enable self-improvement from execution history.
Current video models succeed on basic understanding but achieve under 25% success on logically grounded generation and near 0% on interactive generation, exposing gaps in multimodal reasoning.
ContraPrompt extracts optimization rules from dyadic differences in reasoning traces on identical inputs and organizes them into input-aware decision trees, outperforming GEPA on four benchmarks with gains up to 8.29 pp.
SOCIA-EVO generates statistically consistent simulators by separating structural refinement from parameter calibration via bi-level optimization and falsifying strategies through execution feedback in a Bayesian-weighted playbook.
POES frames prompt evaluation as online adaptive testing and uses a provably submodular objective to pick informative examples, delivering 6.2% higher average accuracy and 35-60% token savings versus naive full-set scoring.
Pioneer Agent automates the full lifecycle of adapting and continually improving small language models via diagnosis-driven data synthesis and regression-constrained retraining, delivering gains of 1.6-83.8 points on benchmarks and large lifts in production-style tasks.
Cognitive Fabric Nodes middleware improves multi-agent LLM system performance by over 10% on HotPotQA and MuSiQue datasets by elevating memory to an active substrate for topology selection, semantic grounding, security, and prompt transformation.
Reflective Context Learning unifies context optimization for agents by recasting prior methods as instances of a shared learning problem and extending them with classical primitives such as batching, failure replay, and grouped rollouts, yielding improvements on AppWorld, BrowseComp+, and RewardBene
Multi-agent deep research systems self-optimize prompts through self-play to match or outperform expert-crafted versions.
UNBOX recovers interpretable text concepts that maximally activate classes in black-box vision models by recasting activation maximization as semantic search with LLMs and diffusion models.
citing papers explorer
-
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.
-
The Moltbook Files: A Harmless Slopocalypse or Humanity's Last Experiment
An AI-agent social platform generated mostly neutral content whose use in fine-tuning reduced model truthfulness comparably to human Reddit data, suggesting limited unique harm but flagging tail risks like secret leaks.
-
More Is Not Always Better: Cross-Component Interference in LLM Agent Scaffolding
Full factorial testing of five LLM agent components reveals that the complete 'All-In' combination is consistently outperformed by smaller subsets due to cross-component interference, with optimal subsets being task- and scale-dependent.
-
TSCG: Deterministic Tool-Schema Compilation for Agentic LLM Deployments
TSCG compiles JSON tool schemas into token-efficient structured text, raising tool-use accuracy for small LLMs from 0% to 84.4% on benchmarks while cutting tokens by 52-57%.
-
RosettaSearch: Multi-Objective Inference-Time Search for Protein Sequence Design
RosettaSearch applies LLM-driven multi-objective search at inference time to improve backbone-conditioned protein sequences, recovering designs with 18-68% better structural fidelity and 2.5x higher success rates than single-pass models like LigandMPNN.
-
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.
-
FVRuleLearner: Operator-Level Reasoning Tree (OP-Tree)-Based Rules Learning for Formal Verification
FVRuleLearner introduces an Operator Reasoning Tree to learn operator-specific rules that improve natural-language to SystemVerilog assertion generation, raising syntax correctness by 3.95% and functional correctness by 31.17% over baselines.
-
Inference-Time Scaling of Verification: Self-Evolving Deep Research Agents via Test-Time Rubric-Guided Verification
DeepVerifier enables self-evolving deep research agents via rubric-guided verification at test time, delivering 8-11% accuracy gains on GAIA and XBench-DeepSearch subsets.
-
Automated Design of Agentic Systems
Meta Agent Search uses a meta-agent to iteratively program novel agentic systems in code, producing agents that outperform state-of-the-art hand-designed ones across coding, science, and math while transferring across domains and models.
-
PIVOT: Bridging Planning and Execution in LLM Agents via Trajectory Refinement
PIVOT refines LLM agent trajectories through plan-inspect-evolve-verify stages using environment feedback, yielding up to 94% relative gains in constraint satisfaction and 3-5x token efficiency over prior refinement methods.
-
MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks
MT-JailBench is a modular benchmark that standardizes evaluation of multi-turn jailbreaks to identify key success drivers and enable stronger combined attacks.
-
AgentSlimming: Towards Efficient and Cost-Aware Multi-Agent Systems
AgentSlimming compresses graph-structured multi-agent systems by estimating agent importance and removing or replacing low-value agents, cutting token costs by up to 78.9% with negligible performance loss.
-
FlashEvolve: Accelerating Agent Self-Evolution with Asynchronous Stage Orchestration
FlashEvolve accelerates LLM agent self-evolution via asynchronous stage orchestration and inspectable language-space staleness handling, reporting 3.5-4.9x proposal throughput gains over synchronous baselines on GEPA workloads.
-
Learning to Evolve: A Self-Improving Framework for Multi-Agent Systems via Textual Parameter Graph Optimization
TPGO represents multi-agent systems as graphs of textual parameters and applies group relative optimization to enable self-improvement from execution history.
-
How Far Are Video Models from True Multimodal Reasoning?
Current video models succeed on basic understanding but achieve under 25% success on logically grounded generation and near 0% on interactive generation, exposing gaps in multimodal reasoning.
-
ContraPrompt: Contrastive Prompt Optimization via Dyadic Reasoning Trace Analysis
ContraPrompt extracts optimization rules from dyadic differences in reasoning traces on identical inputs and organizes them into input-aware decision trees, outperforming GEPA on four benchmarks with gains up to 8.29 pp.
-
SOCIA-EVO: Automated Simulator Construction via Dual-Anchored Bi-Level Optimization
SOCIA-EVO generates statistically consistent simulators by separating structural refinement from parameter calibration via bi-level optimization and falsifying strategies through execution feedback in a Bayesian-weighted playbook.
-
Select Smarter, Not More: Prompt-Aware Evaluation Scheduling with Submodular Guarantees
POES frames prompt evaluation as online adaptive testing and uses a provably submodular objective to pick informative examples, delivering 6.2% higher average accuracy and 35-60% token savings versus naive full-set scoring.
-
Pioneer Agent: Continual Improvement of Small Language Models in Production
Pioneer Agent automates the full lifecycle of adapting and continually improving small language models via diagnosis-driven data synthesis and regression-constrained retraining, delivering gains of 1.6-83.8 points on benchmarks and large lifts in production-style tasks.
-
Scaling Multi-agent Systems: A Smart Middleware for Improving Agent Interactions
Cognitive Fabric Nodes middleware improves multi-agent LLM system performance by over 10% on HotPotQA and MuSiQue datasets by elevating memory to an active substrate for topology selection, semantic grounding, security, and prompt transformation.
-
Reflective Context Learning: Studying the Optimization Primitives of Context Space
Reflective Context Learning unifies context optimization for agents by recasting prior methods as instances of a shared learning problem and extending them with classical primitives such as batching, failure replay, and grouped rollouts, yielding improvements on AppWorld, BrowseComp+, and RewardBene
-
Self-Optimizing Multi-Agent Systems for Deep Research
Multi-agent deep research systems self-optimize prompts through self-play to match or outperform expert-crafted versions.
-
UNBOX: Unveiling Black-box visual models with Natural-language
UNBOX recovers interpretable text concepts that maximally activate classes in black-box vision models by recasting activation maximization as semantic search with LLMs and diffusion models.
-
Reasoning as Gradient: Scaling MLE Agents Beyond Tree Search
Gome reaches 35.1% any-medal rate on MLE-Bench by mapping reasoning to gradient-based updates, outperforming tree search once models are sufficiently capable.
-
Memory in the Age of AI Agents
The paper maps agent memory research via three forms (token-level, parametric, latent), three functions (factual, experiential, working), and dynamics of formation/evolution/retrieval, plus benchmarks and future directions.
-
Agentic Learner with Grow-and-Refine Multimodal Semantic Memory
ViLoMem is a dual-stream grow-and-refine memory system that separates visual and logical error patterns in MLLMs to improve pass@1 accuracy and reduce repeated mistakes across six multimodal benchmarks.
-
Beyond "I cannot fulfill this request": Alleviating Rigid Rejection in LLMs via Label Enhancement
LANCE applies variational inference for label enhancement across multiple rejection categories, supplying gradients to a refinement model that produces safe, non-rigid responses from LLMs.
-
Automated Instruction Revision (AIR): A Structured Comparison of Task Adaptation Strategies for LLM
AIR excels on label-remapping classification tasks while KNN retrieval leads on closed-book QA and fine-tuning leads on structured extraction and event-order reasoning, showing task-dependent adaptation performance.
-
Small Language Model Helps Resolve Semantic Ambiguity of LLM Prompt
A small language model resolves semantic risks and conflicts in prompts via multi-perspective consistency checks, yielding a 2.5-point gain in LLM reasoning performance at $0.02 cost.
-
Supplement Generation Training for Enhancing Agentic Task Performance
SGT trains a lightweight model to generate task-specific supplemental text that improves performance of a larger frozen LLM on agentic tasks without modifying the large model.
-
Statistical Software Engineering with Tuned Variables
AI system maintenance requires treating configuration choices as versioned governed tuned variables promoted via statistical evidence from sampled evaluations.
-
Skills-Coach: A Self-Evolving Skill Optimizer via Training-Free GRPO
Skills-Coach optimizes LLM agent skills via task generation, prompt/code tuning, comparative execution, and traceable evaluation, reporting gains on a 48-skill benchmark called Skill-X.