ModelRefs / Catastrophic Forgetting — AI Glossary
Catastrophic Forgetting — AI Glossary
The tendency of a neural network to abruptly lose previously learned knowledge when fine-tuned on a new task. Critical concern for continual learning pipelines.
Overview
Fine-tuning a model on narrow task data overwrites weights that encoded general capabilities, degrading performance on original tasks. Mitigations: LoRA (update small adapters, not full weights), elastic weight consolidation (EWC), mixture-of-experts routing, and rehearsal (mixing original data). Critical concern for continual learning pipelines.
Reference details
| Topic | training |
|---|---|
| Also known as | catastrophic interference |
| Last reviewed | 2026-06-24 |
Related terms
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Catastrophic Forgetting — AI Glossary.
Frequently asked questions
What is Catastrophic Forgetting?
The tendency of a neural network to abruptly lose previously learned knowledge when fine-tuned on a new task.
Is Catastrophic Forgetting the same as catastrophic interference?
Yes — catastrophic interference are common aliases for Catastrophic Forgetting.
What concepts are related to Catastrophic Forgetting?
Closely related concepts include continual learning, adapter, lora.