SOURCE-LINKED INTELLIGENCE

Replay on Demand: An Emergent Curriculum for Balancing Adaptation and Forgetting in Continued Pretraining

arXiv · AI, language, vision and robotics · article · Sep 30, 2026 · UTC

Continued pretraining enables language models to adapt to new domains and knowledge, but often at the cost of forgetting previously acquired capabilities. Replay can mitigate this trade-off, but fixed replay mixtures allocate training independently of the model's actual retention needs. We introduce Replay on Demand (RoD), which instead derives the replay allocation from the model's learning dynamics. RoD jointly prioritizes adaptation samples by their remaining learning potential and replay samples by their observed forgetting. Their competition for a shared training budget yields an online c

Read original source ↗ Open in workspace

recordType

paper

region

Global