CAMR (Compatibility-Aware Memory Replay) is an adaptive framework for continual learning in large language models that jointly manages supervision of new examples and replay of historical data. It uses compatibility checks and a learned memory gate to decide among three actions: general replay of incoming examples, reinforcing compatible related examples, and skipping supervision for mastered duplicates. Evaluated on Qwen3-8B-Base, CAMR outperforms existing methods on WikiBigEdit factual updates and achieves top scores on multiple TRACE tasks while preserving earlier-task performance across diverse domains.
arXiv:2510. 21978v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has delivered impressive gains in mathematical and multimodal reasoning and has become a standard post-training paradigm for contemporary language and vision-language models.
arXiv:2607. 15587v1 Announce Type: new Abstract: Continual learning studies how deployed language models can continually acquire new tasks without expensive retraining from scratch.
The paper investigates how language models can aid their own continual learning by generating replay data and selecting preservation targets. By organizing task descriptions into Task Banks, the authors show that replay generated by frozen models reduces forgetting on tasks outside the bank and that informed selection of replay targets can improve retention. The study demonstrates that models can support broader self‑preservation, especially when larger models are used, and that replacing new‑task examples with replay can enhance preservation under a fixed example budget.
The paper introduces Golden-GRPO Injection (GRIN), a three-stage self‑learning framework that uses a mixed‑policy reinforcement learning algorithm to inject knowledge into large language models. GRIN injects a golden answer to provide learning signals even when on‑policy rollouts fail on novel facts, and is evaluated on two new document‑level benchmarks—Blank and Counter—that test novel acquisition and counterfactual overwrite. Experiments show that mixed‑policy RL enables knowledge absorption beyond what supervised fine‑tuning can achieve, with GRIN outperforming SFT and other RL baselines on harder question types while matching them on basic fact recall.
arXiv:2603.16157v2 Announce Type: replace-cross Abstract: While Reinforcement Learning (RL) enhances Large Language Model reasoning, on-policy algorithms like GRPO are sample-inefficient as they disc...
The paper introduces Spaced Repetition Training (SRT), a continual learning framework that adapts review scheduling for language models by using the SM-2 algorithm to decide which past examples to replay. SRT tracks per-example review states and maps perplexity to a recall-quality signal, allowing the model to retain old knowledge while consolidating new information without changing the underlying model or training objective. Experiments on Wikipedia and code corpora show that SRT improves the stability-plasticity trade‑off, recovers 5–37 percentage points of lost accuracy, and maintains benchmark performance better than naive continual pre‑training or uniform replay; similar benefits are observed in vision and tabular data when an appropriate recall signal is used.
The paper introduces Spaced Repetition Training (SRT), a continual learning framework that schedules sample rehearsal using the SM-2 algorithm. SRT tracks per-example review states and maps perplexity to recall quality, allowing the training loop to decide which examples to replay and when. Experiments on Wikipedia and code corpora show that SRT improves the stability‑plasticity trade‑off, recovers 5–37 percentage points of lost old‑knowledge accuracy, and preserves benchmark performance better than naive continual pre‑training or uniform replay.
The paper introduces Skill Schema Transport (SST), a post‑training framework that mitigates catastrophic forgetting in continual domain adaptation by translating new domain knowledge into a model’s existing skill space. SST abstracts each example type into a reusable skill schema, preserving essential operations while removing domain‑specific surface forms, thereby turning knowledge updates into procedural reuse rather than trajectory memorization. Experiments on two 3B instruction‑tuned backbones show that SST consistently reduces forgetting and improves generalization, achieving up to a 5.75‑point performance gain and a dramatic drop in maximum forgetting from 42.41 to 7.10 points on the Qwen‑3B MATH stream.
arXiv:2606. 03087v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) improves the ability of large language model, yet headline accuracy gains often conceal a hidden cost: previously solved problems quietly become unsolvable as training proceeds.
arXiv:2609.35954v1 Announce Type: cross Abstract: Large language model post-training generates self-generated rollouts through reinforcement learning and on-policy distillation, yet this experience i...
The paper introduces AMBER, an append‑only memory framework for language‑model agents that interact over long horizons. AMBER lets agents jointly learn to reason, act, and write free‑form memory, guaranteeing retention by construction and enabling end‑to‑end reinforcement learning without extensive curated data. Experiments on WebArena Lite show AMBER outperforms overwrite‑based memory by 4.09 percentage points in average success and improves task completion rates in repeated runs.
arXiv:2610.15715v1 Announce Type: new Abstract: Linear attention promises the efficiency needed for long-context language modelling, yet its fixed-size recurrent state must simultaneously support ong...