self-improving-agents

Tag

Cards List
#self-improving-agents

@dair_ai: Self-improving agents are one of the most discussed ideas in AI right now. (bookmark it) This update gate for a persona…

X AI KOLs Timeline ↗ · yesterday Cached

Safe Personal RSI introduces an update gate for self-improving personal agents that rejects risky self-proposed instruction edits (20 of 21) while allowing safe repairs, cutting attack violations from 10.7% to 0% for a GPT-5.6 Luna repair on the Kicker health assistant.

0 favorites 0 likes
#self-improving-agents

The Resolution Hypothesis: Agent Evolution's Key Is Identifying Improvements

Hugging Face Daily Papers ↗ · 4d ago Cached

The paper introduces the Resolution Hypothesis, arguing that agent evolution stalls because noisy evaluations fail to reliably identify true improvements rather than because proposals are poor. Across seven evolution methods and six models (10M+ LLM calls), it shows candidate selection performs no better than random and published method rankings often fall within run-to-run variance.

0 favorites 0 likes
#self-improving-agents

CompoundingKernel: Self-Improving from Zero to First

Hugging Face Daily Papers ↗ · 2026-11-14 Cached

The paper presents CompoundingKernel, a self-improving agent system that searches for GPU kernels from scratch, ranking #1 on the L1, L2 and Quantization boards of SOL-ExecBench with 4.40x speedups over the reference baseline. It shows that a per-problem kernel archive plus an LLM controller and external correctness checks outperforms a coded UCB-style tree, and that reliable measurement throughput—not agent count—limits scaling.

0 favorites 0 likes
#self-improving-agents

@LangChain: What does it take to get to an agent that catches its own mistake and writes the eval and the fix? Catch @hwchase17's I…

X AI KOLs Following ↗ · 2026-10-27 Cached

LangChain CEO Harrison Chase's Interrupt London session breaks agents into model, harness, and context, showing how each improves via LangSmith Fine-Tuning, LangSmith Engine, and memory tools—closing with a North Star vision of an agent that catches its own mistakes, writes its own evals, and proposes its own fixes.

0 favorites 0 likes
#self-improving-agents

GEPA-Curator: Curated Learning for Self Improving Agents

Hugging Face Daily Papers ↗ · 2026-10-27 Cached

GEPA-Curator is a framework that improves agentic prompt optimization by filtering noisy failure signals and retrieving prior successful trajectories, achieving gains of up to 20.7% over GEPA and 21.2% over ACE on benchmarks like GoBrowse, ALFWorld, AppWorld, and τ-bench.

0 favorites 0 likes
#self-improving-agents

RSIHub: A Composable Framework for Agent-Evolution Methods

Papers with Code Trending ↗ · 2026-10-23 Cached

RSIHub 是一个开源框架,为智能体进化(agent-evolution)研究提供可组合的共享基础设施,标准化方法的构建、比较与扩展,并在 Terminal Bench 2 和 τ3 Banking 上评测了 AHE、Hyperagents、A-Evolve 和 GEPA 等方法。

0 favorites 0 likes
#self-improving-agents

A Self-Improving Agent that Grows its Own DSL: Oracle-Free Operator Induction for Exact Question Answering

arXiv cs.AI ↗ · 2026-10-16 Cached

提出一个无奖励、无裁判、无答案键的自改进代理框架,通过双层循环在小型精确DSL上进行算子归纳,实现精确问答能力的自我扩展,并在多个领域验证其可靠性与跨模型迁移能力。

0 favorites 0 likes
#self-improving-agents

@rohanpaul_ai: Self-improving agents should accept a harness edit only after it beats the current version on fresh, independent tasks.…

X AI KOLs Timeline ↗ · 2026-10-15 Cached

该推文介绍一项关于自我改进智能体的研究,将 harness 自改方法映射到进化算法,并提出准入规则:只有在全新独立任务上以一定幅度超越当前版本时才接受编辑,以防止过拟合与技能遗忘。

0 favorites 0 likes
#self-improving-agents

@rohanpaul_ai: Self-improving agents have always needed a human to write the grader. A new AWS + HSBC paper removes that step: the gra…

X AI KOLs Timeline ↗ · 2026-10-12 Cached

AWS and HSBC researchers present Double Ratchet, a method where a grader evolves alongside a self-improving agent from just 10 labeled examples, matching ground-truth supervision across code, SQL, and report generation tasks.

0 favorites 0 likes
#self-improving-agents

@alacheng: Recursive Self-Improvement (RSI) is the hottest direction in the AI field. In the past six months, many companies have …

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

推文介绍了一个关于递归自我改进(RSI)的精选研究合集,涵盖从STaR、Self-Rewarding LMs到Promptbreeder和Automated Design of Agentic Systems等关键论文,梳理了RSI从自生成训练到自我修改代码代理的演进脉络。

0 favorites 0 likes
#self-improving-agents

@omarsar0: Recommended read. And I agree that the RSI is also a systems engineering problem. Self-improving agents need better res…

X AI KOLs Following ↗ · 2026-10-08 Cached

推荐介绍 RSIGym,将递归式自我改进(RSI)视为系统工程问题,为研究智能体提供训练、推理、评测与沙箱即服务,使 Opus 5 在 SWE-bench Verified 上从 17.67% 提升到 50.33%。

0 favorites 0 likes
#self-improving-agents

Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer

arXiv cs.AI ↗ · 2026-10-01 Cached

提出一种自进化 agent harness 框架:同一冻结模型先作为 solver 解题、再作为 proposer 直接编辑自己的 harness 代码,在多任务上进化后于分布外基准上显著超越 Codex(提升 12.64 分)并达成匹配表现。

0 favorites 0 likes
#self-improving-agents

MERID: Multimodal Exploration via Recursive Self-Improvement Agents for Major Depression Analysis

arXiv cs.AI ↗ · 2026-09-30 Cached

MERID 是一个多模态抑郁症分析框架,通过经验驱动的递归自我改进智能体自动探索和优化检测流水线,在多个抑郁症基准上取得领先结果。

0 favorites 0 likes
#self-improving-agents

AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks

Hugging Face Daily Papers ↗ · 2026-09-29 Cached

AREX-2 advances LLM agent self-improvement by training on synthesized long-horizon reflective trajectories from ML and algorithmic programming tasks. Built on Qwen3.8-27B, it achieves strong results on MLE-bench Lite (81.8) and Frontier-CS (70.7) and transfers to deep research tasks like BrowseComp, HLE, GAIA, and DeepSearchQA.

0 favorites 0 likes
#self-improving-agents

@rohanpaul_ai: New Stanford+Oxford paper MedRSI shows that medical agents can improve themselves from their own mistakes, but only if …

X AI KOLs Following ↗ · 2026-09-23 Cached

A Stanford+Oxford paper on MedRSI shows medical AI agents can self-improve from mistakes by testing new capabilities on fresh patients and using guardrails to maintain accuracy.

0 favorites 0 likes
#self-improving-agents

@arunmoorthy05: https://x.com/arunmoorthy05/status/2099936495465631943

X AI KOLs Timeline ↗ · 2026-09-15 Cached

Pace details their approach to improving an AI extraction agent for insurance documents by optimizing subagent delegation, model routing, and context based on historical data, leading to reduced cost and latency.

0 favorites 0 likes
#self-improving-agents

@Saboo_Shubham_: Stanford dropped a full course on self-improving AI agents. 100% free.

X AI KOLs Following ↗ · 2026-09-07 Cached

Stanford University has launched a complete free course on self-improving AI agents, providing accessible educational resources for AI advancements.

0 favorites 0 likes
#self-improving-agents

@rohanpaul_ai: Self-improving agents need a useful way to remember what worked. SkillGLoW shows agents should remember reusable ways o…

X AI KOLs Timeline ↗ · 2026-09-05 Cached

SkillGLoW demonstrates that self-improving agents achieve better performance by storing reusable task-solving procedures rather than memorizing every past task, gaining 17.2 points with a 3.6× more compact library.

0 favorites 0 likes
#self-improving-agents

LLM-as-a-Judge Is Not an Oracle: Why Self-Improving Agents Need Deterministic Guardrails

arXiv cs.AI ↗ · 2026-09-03 Cached

This paper identifies failure modes in LLM-as-a-Judge systems for self-improving agents and introduces PROCTOR, an architecture with deterministic guardrails to mitigate these issues.

0 favorites 0 likes
#self-improving-agents

SkillGLoW: Procedural-Family Skill Consolidation for Self-Improving Agents on Long-Horizon Task Streams

arXiv cs.AI ↗ · 2026-09-03 Cached

SkillGLoW is a method for LLM agents that organizes skills into procedural families to enhance self-improvement on long-horizon tasks, demonstrating significant performance gains over baselines.

0 favorites 0 likes
Next →
← Back to home

Submit Feedback