Memento: Fine-tuning LLM Agents without Fine-tuning LLMs
read the original abstract
In this paper, we introduce a novel learning paradigm for Adaptive Large Language Model (LLM) agents that eliminates the need for fine-tuning the underlying LLMs. Existing approaches are often either rigid, relying on static, handcrafted reflection workflows, or computationally intensive, requiring gradient updates of LLM model parameters. In contrast, our method enables low-cost continual adaptation via memory-based online reinforcement learning. We formalise this as a Memory-augmented Markov Decision Process (M-MDP), equipped with a neural case-selection policy to guide action decisions. Past experiences are stored in an episodic memory, either differentiable or non-parametric. The policy is continually updated based on environmental feedback through a memory rewriting mechanism, whereas policy improvement is achieved through efficient memory reading (retrieval). We instantiate our agent model in the deep research setting, namely \emph{Memento}, which attains top-1 on GAIA validation ($87.88\%$ Pass@$3$) and $79.40\%$ on the test set. It reaches $66.6\%$ F1 and $80.4\%$ PM on the DeepResearcher dataset, outperforming the state-of-the-art training-based method, while case-based memory adds $4.7\%$ to $9.6\%$ absolute points on out-of-distribution tasks. Our approach offers a scalable and efficient pathway for developing generalist LLM agents capable of continuous, real-time learning without gradient updates, advancing machine learning towards open-ended skill acquisition and deep research scenarios. The code is available at https://github.com/Agent-on-the-Fly/Memento.
This paper has not been read by Pith yet.
Forward citations
Cited by 40 Pith papers
-
Metadata, Structure, or Strategy? A Decomposition of RAG Context Enrichment
Controlled experiments across six benchmarks and four models show RAG context enrichment with metadata, structure, or strategies mostly lowers accuracy, with model-context alignment as the determining factor.
-
Experience Makes Skillful: Enabling Generalizable Medical Agent Reasoning via Self-Evolving Skill Memory
SkeMex distills agent trajectories into value-aware skills organized in general/task/action branches and evolves them via a closed-loop Read-Write-Assess-Govern process, outperforming prior memory agents on clinical tasks.
-
Co-Evolving Skill Generation and Policy Optimization
Framework estimates context-dependent marginal utility of candidate skills via reward gaps in matched base vs. skill-augmented rollouts to filter skills and co-train policy as generator.
-
Rosetta Memory: Adaptive Memory for Cross-LLM Agents
Rosetta Memory trains two profile-conditioned operators with a minimum-gain sampling curriculum and performance-gap reward to enable memory transfer between LLMs, showing gains on multi-hop QA benchmarks and robustnes...
-
Autonomous heterogeneous catalyst discovery with a self-evolving multi-agent digital twin
CatDT deploys a self-evolving multi-agent system with UniMech and reinforcement learning to build digital twins of heterogeneous catalysts, matching experimental rates within 0.5-2x on seven benchmarks and identifying...
-
OLIVIA: Online Learning via Inference-time Action Adaptation for Decision Making in LLM ReAct Agents
OLIVIA treats LLM agent action selection as a contextual linear bandit over frozen hidden states and applies UCB exploration to adapt online, yielding consistent gains over static ReAct and prompt-based baselines on f...
-
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.
-
From Skills to Talent: Organising Heterogeneous Agents as a Real-World Company
OMC framework turns multi-agent AI into self-organizing companies with Talents, Talent Market, and E²R search, achieving 84.67% success on PRDBench (15.48 points above prior art).
-
Springdrift: An Auditable Persistent Runtime for LLM Agents with Case-Based Memory, Normative Safety, and Ambient Self-Perception
Springdrift provides an auditable persistent runtime for long-lived LLM agents with case-based memory, normative safety gating, and ambient self-perception, shown in a 23-day single-instance deployment where the agent...
-
Metis: Bridging Text and Code Memory for Self-Evolving Agents
Metis combines text and code memory hierarchically for self-evolving agents, claiming up to 20.6% higher accuracy and 22.8% lower cost than ReAct on the AppWorld benchmark.
-
HarnessX: A Composable, Adaptive, and Evolvable Agent Harness Foundry
HarnessX assembles and evolves agent harnesses via substitution algebra and AEGIS trace analysis, reporting +14.5% average gains (up to +44%) on five benchmarks.
-
Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agents
HORMA builds a hierarchical memory structure from agent experiences and trains a lightweight RL navigator to retrieve minimal sufficient context, yielding better task performance with at most 22.17% of baseline token ...
-
From Player to Master: Enhancing Test-Time Learning of LLM Agents via Reinforcement Learning over Memory
MemoPilot trains memory updates for LLM agents via multi-turn GRPO on RPS and poker, achieving top Elo scores and outperforming baselines including DeepSeek-V3.2.
-
Traj-Evolve: A Self-Evolving Multi-Agent System for Patient Trajectory Modeling in Lung Cancer Early Detection
Traj-Evolve combines non-parametric experience retrieval and multi-agent RL with a leave-one-out unification strategy to outperform baselines on lung cancer prediction from up to five years of multimodal EHRs, includi...
-
Rethinking Memory as Continuously Evolving Connectivity
FluxMem evolves memory as a heterogeneous graph via three refinement stages and reports consistent state-of-the-art results on LoCoMo, Mind2Web, and GAIA benchmarks.
-
Mem-$\pi$: Adaptive Memory through Learning When and What to Generate
Mem-π is a framework using a dedicated model and decision-content decoupled RL to generate context-specific guidance on demand for LLM agents, outperforming retrieval baselines by over 30% on web navigation.
-
EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective
EvoMemBench evaluates 15 memory methods for LLM agents and finds long-context baselines competitive with no single memory approach working consistently across settings.
-
Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM Memory
SeqMem-Eval reveals that high final accuracy in sequential LLM memory tasks often coexists with substantial forgetting and negative transfer, exposing stability-adaptability trade-offs hidden by standard aggregate metrics.
-
PREPING: Building Agent Memory without Tasks
Preping builds agent memory via proposer-guided synthetic practice and selective validation, matching offline/online methods at 2-3x lower deployment cost.
-
Skill-R1: Agent Skill Evolution via Reinforcement Learning
Skill-R1 applies bi-level group-relative policy optimization to evolve skills recurrently from verified outcomes, yielding gains over baselines on multi-step tasks.
-
Learning Agent Routing From Early Experience
BoundaryRouter routes queries to LLM or agent using early experience memory from a seed set, cutting inference time 60.6% versus always using agents and raising performance 28.6% versus always using direct LLM inference.
-
Skill1: Unified Evolution of Skill-Augmented Agents via Reinforcement Learning
Skill1 trains one policy to jointly evolve skill query generation, re-ranking, task solving, and distillation from a single task-success signal, with low-frequency trends crediting selection and high-frequency variati...
-
CASCADE: Case-Based Continual Adaptation for Large Language Models During Deployment
CASCADE enables LLMs to continually adapt at deployment via case-based episodic memory and contextual bandits, improving macro-averaged success by 20.9% over zero-shot on 16 tasks spanning medicine, law, code, and robotics.
-
StepPO: Step-Aligned Policy Optimization for Agentic Reinforcement Learning
StepPO reformulates agentic RL as a step-level MDP with step-level credit assignment and importance sampling, consistently outperforming token-level and trajectory-level baselines across four agent benchmarks.
-
MEMENTO: Teaching LLMs to Manage Their Own Context
MEMENTO trains LLMs to segment reasoning into blocks, generate mementos as dense summaries, and reason forward using only mementos and KV states, cutting peak KV cache by ~2.5x while preserving benchmark accuracy.
-
EvolveR: Self-Evolving LLM Agents through an Experience-Driven Lifecycle
EvolveR enables LLM agents to self-evolve via a closed loop of distilling interaction trajectories into strategic principles offline and retrieving them to guide online decisions with policy reinforcement, yielding be...
-
Procedural Memory Distillation: Online Reflection for Self-Improving Language Models
PMD extracts and distills cross-episode procedural knowledge from RL rollouts into LLM policies at three abstraction levels, yielding 3.8-13.6% gains over SDPO on SCIKNOWEVAL and LIVECODEBENCH via co-evolution.
-
The Past Is Prologue: A Plug-in Controller for Selective Updates in Sequentially Evolving LLM Memory
Janus is a method-agnostic plug-in that uses a Memory Momentum Trigger and compact hybrid evaluation to selectively accept LLM memory updates, yielding +2.7 to +4.6 accuracy gains over base updaters on six datasets.
-
Towards Persistent Case-Based Memory for Autonomous Data Science: A CBR-Augmented R&D-Agent with a Locally Deployable Small Language Model
CBR integration into R&D-Agent with Gemma 4 31B yields directionally higher accuracy and lower variance than baseline on one of two Kaggle competitions.
-
Dynamic Mixture of Latent Memories for Self-Evolving Agents
MoLEM achieves a 10.40% average accuracy improvement in continual learning tasks across math, science, and code by using dynamic latent memory experts with a frozen base model and stage-specific autoencoders for routing.
-
Memory-Augmented Reinforcement Learning Agent for CAD Generation
Memory-augmented RL agent with case and skill libraries plus dynamic retrieval improves success rate and geometric consistency for complex CAD model generation.
-
SkillsVote: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution
SkillsVote is a governance system for agent skills that profiles corpora, recommends via search, and gates updates on successful reusable outcomes, yielding benchmark gains without model changes.
-
Improving Multi-turn Dialogue Consistency with Self-Recall Thinking
SRT framework improves multi-turn dialogue F1 by 4.7% and cuts end-to-end latency by 14.7% via dependency construction, capability initialization, and reasoning improvement with recall tokens.
-
Ace-Skill: Bootstrapping Multimodal Agents with Prioritized and Clustered Evolution
Ace-Skill boosts multimodal agent self-evolution via prioritized rollouts with lazy-decay tracking and semantic knowledge clustering, yielding up to 35% relative gains on tool-use benchmarks and zero-shot transfer to ...
-
Skill1: Unified Evolution of Skill-Augmented Agents via Reinforcement Learning
Skill1 trains a single RL policy to co-evolve skill selection, utilization, and distillation in language model agents from one task-outcome reward, using low-frequency trends to credit selection and high-frequency var...
-
Skill1: Unified Evolution of Skill-Augmented Agents via Reinforcement Learning
Skill1 co-evolves skill selection, utilization, and distillation inside a single policy using only task-outcome reward, with low-frequency trends crediting selection and high-frequency variation crediting distillation...
-
Externalization in LLM Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering
LLM agent progress depends on externalizing cognitive functions into memory, skills, protocols, and harness engineering that coordinates them reliably.
-
SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation
SOLAR introduces a self-optimizing agent using meta-learning on model weights and RL-driven strategy discovery for lifelong adaptation in LLMs, claiming superior performance on reasoning tasks across domains.
-
EvolveR: Self-Evolving LLM Agents through an Experience-Driven Lifecycle
EvolveR proposes a closed-loop self-evolution system for LLM agents that distills experiences into principles offline and applies reinforcement during online task interactions to achieve better performance on multi-ho...
-
StepPO: Step-Aligned Policy Optimization for Agentic Reinforcement Learning
StepPO argues that LLM agents should optimize at the step level rather than token level to better handle delayed rewards and long contexts in agentic RL.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.