self-evolution

Tag

Cards List
#self-evolution

Space-X: Self-Evolving Spatial Agents Weaving Clues into Evidence Chains

arXiv cs.AI ↗ · yesterday Cached

Space-X is a self-evolving spatial-reasoning framework where an Execution Agent builds evidence chains with role-specific memory and a Control Agent diagnoses breaks to jointly revise tools, skills, and decision rules, yielding large accuracy gains across benchmarks with four base models.

0 favorites 0 likes
#self-evolution

@mattpocockuk: This might be the greatest issue report in history Some folks wrote a paper about my /retro skill assessing its effecti…

X AI KOLs Timeline ↗ · 2026-11-03 Cached

An empirical arXiv study examines /retro, a skill that lets coding agents review completed sessions and propose harness changes, finding that while many proposals are generated, only some reduce repeated work on later tasks — showing harness self-evolution needs mechanisms to verify and revise changes.

0 favorites 0 likes
#self-evolution

FilingLoop: Evaluating and Evolving LLM Agents for Corporate Disclosure Review and Revision

arXiv cs.AI ↗ · 2026-10-31 Cached

FilingLoop introduces an interactive benchmark and self-evolving LLM agent framework for corporate disclosure review and revision, grounded in 380 SEC IPO review tasks with requirement-level evaluation that improves fulfillment from 24.28% to 27.96% via procedural learning without weight updates.

0 favorites 0 likes
#self-evolution

PhysEvolve: Verifier-grounded Self-evolution for Physical Reasoning with Executable Worlds

Hugging Face Daily Papers ↗ · 2026-10-26 Cached

PhysEvolve 提出一个验证器驱动的自演化框架,利用物理引擎验证模型自生成的可执行世界与反事实仿真,以 GRPO 训练视觉语言模型的因果物理推理能力,在 PhysBench、CausalPhys 和 ContPhy 上取得显著提升且无需人类干预数据。

0 favorites 0 likes
#self-evolution

@snwiki238337: Current Harness optimization has two main directions. One is a unified global scaffold trying to use a single set of prompts and control logic. But as business scenarios grow, instructions optimized for financial spreadsheets often interfere with code generation tasks, causing severe rule conflicts and excessive context bloat. The other is to copy a separate specialized codebase for each specific domain...

X AI KOLs Following ↗ · 2026-10-22 Cached

Google Cloud AI Research, in collaboration with UIUC, proposes OO-Harness, which introduces object-oriented design (inheritance, encapsulation, polymorphism) into the automatic evolution of agent harnesses. Through a class hierarchy, it enables code reuse and safe fallback for specialized scaffolds. On three long-horizon benchmarks, the main metric improves from 61.62% to 77.41%, and it significantly outperforms traditional self-evolution methods on out-of-distribution tasks.

0 favorites 0 likes
#self-evolution

VibeGame: Prompt-to-Game Development with AI-Native Engine and Self-Evolving Adversarial Agent Team

Hugging Face Daily Papers ↗ · 2026-10-22 Cached

VibeGame is a framework that builds complete 2D games from natural-language prompts using an AI-native game engine with frame-synchronous runtime control and an eight-agent adversarial team that separates implementation from review. It introduces a training-free self-evolution mechanism distilling accepted projects into reusable components, outperforming other agent harnesses on the GameCraft-Bench subset.

0 favorites 0 likes
#self-evolution

Harness Self-Evolution for Long-Horizon Terminal Agents

arXiv cs.AI ↗ · 2026-10-21 Cached

The paper presents RiseHarness, a framework from Tencent and multiple universities that iteratively improves agent harnesses for long-horizon terminal tasks while keeping model weights fixed, analyzing successful and failed runs to progressively integrate better execution mechanisms.

0 favorites 0 likes
#self-evolution

Evolving with Open Worlds: Towards Recursive Self-Improvement for Future Forecasting

Hugging Face Daily Papers ↗ · 2026-10-21 Cached

论文提出 DeepEvolve,一个基于未来事件预测的可持续进化智能体系统,通过在动态开放环境中与现实世界结果对齐实现递归自我改进(RSI),其自我改进的预测器在 FutureX 排行榜连续三周排名第一,评级提升最高达 195%。

0 favorites 0 likes
#self-evolution

From VLM Planners to Robotic Agents: Self-Evolving Harnesses for History-Dependent Manipulation

Hugging Face Daily Papers ↗ · 2026-10-21 Cached

论文提出 SERA(Self-Evolving Robotic Agent),通过可编辑的 harness 维护任务上下文并利用执行反馈自我进化规划,配合固定的 VLA 执行器来改进长时程机器人操作任务,在 RoboMME 上取得 51.38% 的成功率。

0 favorites 0 likes
#self-evolution

Reward Engine: Reward-Centric Self-Evolution for Open-Ended Tool Agents

arXiv cs.AI ↗ · 2026-10-19 Cached

华为openJiuwen团队提出Reward Engine,通过构建由测试用例和评分细则组成的冻结RewardPack,为开放式工具智能体提供证据驱动的奖励信号,用于自修订与强化学习,在WorkBuddy Bench Office和Claw-Eval上显著提升通过一致性。

0 favorites 0 likes
#self-evolution

EvolvingBrain: An Adaptive Skill System for Multi-modal Agents Self-Evolution in Open Worlds

arXiv cs.AI ↗ · 2026-10-16 Cached

提出了EvolvingBrain,一个用于开放世界多模态智能体自我进化的自适应技能系统,通过分层技能自进化和状态感知自适应执行,在VAB-Minecraft和DiscoveryWorld上提升了任务成功率与泛化能力。

0 favorites 0 likes
#self-evolution

@shao__meng: Huawei has open-sourced a multi-agent collaboration framework called "JiuwenSwarm" The core philosophy is "Understand I…

X AI KOLs Timeline ↗ · 2026-10-11 Cached

华为开源了多智能体协作框架「九问Swarm」(JiuwenSwarm),采用无内嵌模型的纯Harness设计,支持E2A统一协议、Leader/Teammate分布式执行、确定性工作流SwarmFlow,以及带人工审核的技能自进化与AutoHarness机制。

0 favorites 0 likes
#self-evolution

The Harness as the Only Mutable Surface: Compliance-Bounded Self-Evolution of LLM Agents in Credit Pipelines, with a Measured Admission Gate

arXiv cs.AI ↗ · 2026-10-09 Cached

该论文提出一种将LLM代理的自进化限制在运行时harness(指令文本、工具调用逻辑与原语组合)的框架,模型权重保持固定,使每次适应都成为可审查的diff,并通过哈希链记录的准入门进行管控。模拟实验显示受控双循环仅放行144/7449个候选变更且未恶化误差,而无约束系统放行了309个有害变更。

0 favorites 0 likes
#self-evolution

Principles that Guide, Actions that Inform: Agent Evolution via Knowledge Abstraction

arXiv cs.AI ↗ · 2026-10-08 Cached

本文提出 SAGA 框架,通过将交互轨迹逐步抽象为可复用的原则与程序,并在执行中实例化和校验这些原则,形成执行-抽象反馈循环,使 LLM 智能体能在不微调参数的情况下持续从经验中进化,在 ScienceWorld 和 ALFWorld 上取得性能提升。

0 favorites 0 likes
#self-evolution

@Gorden_Sun: MIT Open-Sources JAZ Framework: Ending Complex Agent Add-Ons with Minimalist Invoke (Primitive Call) In building Agents…

X AI KOLs Timeline ↗ · 2026-10-05 Cached

MIT 团队开源了 JAZ 框架,用纯提示和极简的 invoke 循环替代复杂记忆附加系统,在长程召回基准上以更低成本超过 Letta (MemGPT) 和 ACE。

0 favorites 0 likes
#self-evolution

VERSE: Verified Self-Evolving Optimizer for Agent Harnesses

arXiv cs.AI ↗ · 2026-10-05 Cached

The paper introduces VERSE, a verified self-evolving optimizer that improves both the executor agent's harness and its own prompts, tools, and procedures using execution-based verification, achieving 42.3% and 37.7% accuracy on held-out SWE-rebench tasks versus 39.2% and 29.3% for the strongest baselines.

0 favorites 0 likes
#self-evolution

Questioning the Questions: Sustaining Self-Evolution in Reasoning Models

Hugging Face Daily Papers ↗ · 2026-10-03 Cached

The paper identifies why self-evolving reasoning models suffer performance collapse — invalid questions and mathematically-equivalent question repetition — and proposes R-Quest, which uses validity and novelty feedback to sustain self-evolution, outperforming R-Zero by 17.32 points across 12 benchmarks over ten rounds.

0 favorites 0 likes
#self-evolution

It Takes Workflows to Evolve Better Workflows

Hugging Face Daily Papers ↗ · 2026-10-01 Cached

FloWright proposes a hierarchical, structure-aware reward paradigm enabling LLM agents in multi-agent workflows to self-evolve and co-evolve without extra models or labels, while DataWright adaptively hardens datasets into workflow-level tasks. Small open models trained with FloWright gain up to +7.41% performance, with co-evolving multiple roles (+5.03%) outperforming optimizing a single role (+2.83%).

0 favorites 0 likes
#self-evolution

@shao__meng: https://x.com/shao__meng/status/2104096547747291584

X AI KOLs Timeline ↗ · 2026-09-27 Cached

Professor Li Hongyi from National Taiwan University has updated the 2026 spring machine learning course, focusing on AI Agents and model self-evolution in the large model era. This marks a complete transition of the course from classical machine learning to cutting-edge large model technologies.

0 favorites 0 likes
#self-evolution

Program-Verified Self-Evolution for Vision-Language Models

Hugging Face Daily Papers ↗ · 2026-09-27 Cached

This paper introduces VQS, a method for self-evolving vision-language models that uses program verification to generate and label questions from images, significantly improving accuracy and model performance over existing baselines.

0 favorites 0 likes
Next →
← Back to home

Submit Feedback