pith. sign in

arxiv: 2310.10134 · v1 · pith:3ZPK72VYnew · submitted 2023-10-16 · 💻 cs.CL · cs.AI· cs.LG

CLIN: A Continually Learning Language Agent for Rapid Task Adaptation and Generalization

classification 💻 cs.CL cs.AIcs.LG
keywords continuallyagentsclinimproveperformancetasktasksagent
0
0 comments X
read the original abstract

Language agents have shown some ability to interact with an external environment, e.g., a virtual world such as ScienceWorld, to perform complex tasks, e.g., growing a plant, without the startup costs of reinforcement learning. However, despite their zero-shot capabilities, these agents to date do not continually improve over time beyond performance refinement on a specific task. Here we present CLIN, the first language-based agent to achieve this, so that it continually improves over multiple trials, including when both the environment and task are varied, and without requiring parameter updates. Our approach is to use a persistent, dynamic, textual memory centered on causal abstractions (rather than general "helpful hints") that is regularly updated after each trial so that the agent gradually learns useful knowledge for new trials. In the ScienceWorld benchmark, CLIN is able to continually improve on repeated trials on the same task and environment, outperforming state-of-the-art reflective language agents like Reflexion by 23 absolute points. CLIN can also transfer its learning to new environments (or new tasks), improving its zero-shot performance by 4 points (13 for new tasks) and can further improve performance there through continual memory updates, enhancing performance by an additional 17 points (7 for new tasks). This suggests a new architecture for agents built on frozen models that can still continually and rapidly improve over time.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 18 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Training the Orchestrator: A Supervised Approach to End-to-End PDDL Planning with LLM Agents

    cs.AI 2026-06 unverdicted novelty 7.0

    HALO trains an orchestrator policy on verifier-approved refinement trajectories across 11 PDDL domains, matching GPT-5-mini success rates at roughly 45x lower orchestration cost and cutting LLM calls by 40-50%.

  2. SubtleMemory: A Benchmark for Fine-Grained Relational Memory Discrimination in Long-Horizon AI Agents

    cs.AI 2026-06 unverdicted novelty 7.0

    SubtleMemory benchmark with 1,522 instances over 10 histories shows current memory systems are weak at fine-grained relational discrimination in long-term AI agent interactions.

  3. State-Centric Decision Process

    cs.AI 2026-05 unverdicted novelty 7.0

    SDP constructs a task-induced state space from raw text by having agents commit to and certify natural-language predicates as states, enabling structured planning and analysis in unstructured language environments.

  4. MetaPS: Adaptive Programmatic Strategy Selection for Market Agents

    cs.AI 2026-06 unverdicted novelty 6.0

    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.

  5. GitOfThoughts: Version-Controlled Reasoning and Agent Memory You Can Replay, Diff, and Merge

    cs.AI 2026-06 unverdicted novelty 6.0

    GitOfThoughts stores agent reasoning as a git repo and shows memory from past problems improves accuracy only when new problems are nearly identical (cosine similarity >0.8), with self-consistency providing the main g...

  6. SKILL.nb: Selective Formalization and Gated Execution for Durable Agent Workflows

    cs.AI 2026-06 unverdicted novelty 6.0

    SKILL.nb uses selective formalization and gate-conditioned execution in auditable notebooks to improve durability of agent workflows, achieving 53.7% success on WebArena-Verified with 91.7% retention across re-executions.

  7. COMAP: Co-Evolving World Models and Agent Policies for LLM Agents

    cs.AI 2026-06 unverdicted novelty 6.0

    COMAP co-evolves textual world models and agent policies for LLMs through on-policy self-distillation, yielding up to 16.75% relative gains on embodied planning, web navigation, and tool-use tasks.

  8. CogniFold: Always-On Proactive Memory via Cognitive Folding

    cs.AI 2026-05 unverdicted novelty 6.0

    Cognifold is a new proactive memory architecture that folds event streams into emergent cognitive structures by extending complementary learning systems theory with a prefrontal intent layer and graph topology self-or...

  9. When Continual Learning Moves to Memory: A Study of Experience Reuse in LLM Agents

    cs.LG 2026-04 unverdicted novelty 6.0

    External memory does not eliminate continual learning challenges in LLM agents but reshapes them into issues of memory representation and retrieval design, with abstract memories aiding transfer while organization cho...

  10. Real-Time Procedural Learning From Experience for AI Agents

    cs.AI 2025-11 unverdicted novelty 6.0

    PRAXIS enables AI agents to acquire procedural knowledge in real time by indexing and retrieving state-action-result experiences, leading to better accuracy, reliability, and efficiency on web browsing benchmarks.

  11. Agent Workflow Memory

    cs.CL 2024-09 unverdicted novelty 6.0

    AWM induces reusable workflows from agent experiences and provides them selectively to improve success rates by 24.6% on Mind2Web and 51.1% on WebArena while reducing steps taken.

  12. Always-OnAgents:A Survey of Persistent Memory, State, and Governance in LLMAgents

    cs.MA 2026-06 unverdicted novelty 5.0

    Survey mapping persistent state in LLM agents along six axes and proposing the AOEP-v0 protocol to evaluate governance and recovery obligations.

  13. Robo-Cortex: A Self-Evolving Embodied Agent via Dual-Grain Cognitive Memory and Autonomous Knowledge Induction

    cs.RO 2026-05 unverdicted novelty 5.0

    Robo-Cortex proposes a self-evolving embodied navigation agent using dual-grain cognitive memory and autonomous knowledge induction from trajectories, reporting SPL gains on IGNav, AR, AEQA and preliminary real-robot tests.

  14. FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast

    cs.AI 2026-05 unverdicted novelty 5.0

    FORGE is a staged population protocol that evolves prompt-injected memory (Rules, Examples, or Mixed) for ReAct agents via reflection and broadcast, yielding 1.7-7.7× gains over zero-shot and 29-72% over Reflexion on ...

  15. CogniFold: Always-On Proactive Memory via Cognitive Folding

    cs.AI 2026-05 unverdicted novelty 5.0

    CogniFold extends Complementary Learning Systems theory to three layers with a prefrontal intent layer and uses graph self-organization to build proactive agent memory from continuous event streams.

  16. A Control Architecture for Training-Free Memory Use

    cs.AI 2026-04 unverdicted novelty 5.0

    A training-free control architecture with uncertainty-based routing, confidence-selective acceptance, and evidence-based memory governance improves arithmetic reasoning by +7 points on SVAMP and ASDiv benchmarks.

  17. Iterative Formalization and Planning in Partially Observable Environments

    cs.AI 2025-05 unverdicted novelty 5.0

    PDDLego iteratively formalizes and refines PDDL representations of partially observable environments to improve planning success without finetuning or in-context examples.

  18. From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanisms

    cs.AI 2026-05 unverdicted novelty 4.0

    LLM agent memory is organized into Storage (preserving trajectories), Reflection (refining them), and Experience (abstracting into reusable knowledge) stages driven by needs for long-range consistency, dynamic adaptat...