When Does Continual Learning Require Learning

• URL: http://arxiv.org/abs/2607.07847v1
• Date: Wed, 08 Jul 2026 18:27:40 GMT
• Title: When Does Continual Learning Require Learning
• Abstract summary: We show that continual learning is not a single capability.<n>Different patterns of environmental change require fundamentally different update behaviors.<n>We hope that understanding where each method succeeds and fails will guide the design of stronger continual learning systems.
• Score: 68.55508958298942
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As large language models (LLMs) become increasingly capable, the next question is how can we enable models to continually learn? Today, the field largely frames this as a problem of context management and mitigating forgetting. We argue this framing is incomplete: continual learning is fundamentally about increasing model competence as the world changes. We disentangle this change along two axes -- space, where the model encounters new domains, and time, where the underlying data drifts under a fixed task. This framing lets us study continual learning under realistic conditions: new domains arrive over time, facts drift past their training cutoff, and agentic interactions accumulate state across episodes. To evaluate methods under this setting, we recast widely used LLM benchmarks as sequential problems and introduce a single mechanism-agnostic protocol that compares prompt-based methods (GEPA, ACE), supervised learning (SFT, SDFT), reinforcement learning (GRPO, SDPO), and context compression (Cartridges, In-place TTT). Prompt-based methods fit each new stage quickly but degrade on future tasks. Distillation-based methods accumulate knowledge stably but struggle to update outdated facts. Context compression improves efficiency without substantially improving the ability to learn new tasks. Online reinforcement learning adapts most effectively to knowledge updates but remains sensitive to noisy reward signals. Overall, our results suggest that continual learning is not a single capability: different patterns of environmental change require fundamentally different update behaviors, determining when adaptation must be learned inside model weights and when it can be achieved through external scaffolding. We hope that understanding where each method succeeds and fails will guide the design of stronger continual learning systems.

Related papers

• Nested Learning: The Illusion of Deep Learning Architectures [57.41377373511876]
  We present a new learning paradigm, called Nested Learning (NL), that coherently represents a machine learning model with a set of nested, multi-level, and/or parallel problems.<n>We show three core contributions: Expressive generalizations are in fact as generalizations with deep memory and/or more powerful learning rules.<n>We present a new continuum for memory system that generalizes the traditional viewpoint of long/short-term memory.
  arXiv  Detail & Related papers  (2025-12-31T07:59:43Z)
• Tackling Distribution Shift in LLM via KILO: Knowledge-Instructed Learning for Continual Adaptation [0.35297361401370037]
  Large Language Models (LLMs) often suffer from performance degradation when faced with domain shifts.<n>We propose KILO, a novel continual learning framework that integrates dynamic knowledge graphs with instruction tuning.
  arXiv  Detail & Related papers  (2025-08-05T15:39:37Z)
• Analytic Subspace Routing: How Recursive Least Squares Works in Continual Learning of Large Language Model [6.42114585934114]
  Large Language Models (LLMs) possess capabilities that can process diverse language-related tasks.<n>Continual Learning in Large Language Models (LLMs) arises which aims to continually adapt the LLMs to new tasks.<n>This paper proposes Analytic Subspace Routing(ASR) to address these challenges.
  arXiv  Detail & Related papers  (2025-03-17T13:40:46Z)
• Beyond Prompt Learning: Continual Adapter for Efficient Rehearsal-Free Continual Learning [22.13331870720021]
  We propose a beyond prompt learning approach to the RFCL task, called Continual Adapter (C-ADA) C-ADA flexibly extends specific weights in CAL to learn new knowledge for each task and freezes old weights to preserve prior knowledge. Our approach achieves significantly improved performance and training speed, outperforming the current state-of-the-art (SOTA) method.
  arXiv  Detail & Related papers  (2024-07-14T17:40:40Z)
• GrowOVER: How Can LLMs Adapt to Growing Real-World Knowledge? [36.987716816134984]
  We propose GrowOVER-QA and GrowOVER-Dialogue, dynamic open-domain QA and dialogue benchmarks that undergo a continuous cycle of updates. Our research indicates that retrieval-augmented language models (RaLMs) struggle with knowledge that has not been trained on or recently updated. We introduce a novel retrieval-interactive language model framework, where the language model evaluates and reflects on its answers for further re-retrieval.
  arXiv  Detail & Related papers  (2024-06-09T01:16:04Z)
• Can LMs Learn New Entities from Descriptions? Challenges in Propagating Injected Knowledge [72.63368052592004]
  We study LMs' abilities to make inferences based on injected facts (or propagate those facts) We find that existing methods for updating knowledge show little propagation of injected knowledge. Yet, prepending entity definitions in an LM's context improves performance across all settings.
  arXiv  Detail & Related papers  (2023-05-02T17:59:46Z)
• Online Continual Learning via the Knowledge Invariant and Spread-out Properties [4.109784267309124]
  Key challenge in continual learning is catastrophic forgetting. We propose a new method, named Online Continual Learning via the Knowledge Invariant and Spread-out Properties (OCLKISP) We empirically evaluate our proposed method on four popular benchmarks for continual learning: Split CIFAR 100, Split SVHN, Split CUB200 and Split Tiny-Image-Net.
  arXiv  Detail & Related papers  (2023-02-02T04:03:38Z)
• Fully Online Meta-Learning Without Task Boundaries [80.09124768759564]
  We study how meta-learning can be applied to tackle online problems of this nature. We propose a Fully Online Meta-Learning (FOML) algorithm, which does not require any ground truth knowledge about the task boundaries. Our experiments show that FOML was able to learn new tasks faster than the state-of-the-art online learning methods.
  arXiv  Detail & Related papers  (2022-02-01T07:51:24Z)
• Meta-learning the Learning Trends Shared Across Tasks [123.10294801296926]
  Gradient-based meta-learning algorithms excel at quick adaptation to new tasks with limited data. Existing meta-learning approaches only depend on the current task information during the adaptation. We propose a 'Path-aware' model-agnostic meta-learning approach.
  arXiv  Detail & Related papers  (2020-10-19T08:06:47Z)
• Bilevel Continual Learning [76.50127663309604]
  We present a novel framework of continual learning named "Bilevel Continual Learning" (BCL) Our experiments on continual learning benchmarks demonstrate the efficacy of the proposed BCL compared to many state-of-the-art methods.
  arXiv  Detail & Related papers  (2020-07-30T16:00:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.