🚨 Introducing UPCORE, to balance deleting info from LLMs with keeping their other capabilities intact.
UPCORE selects a coreset of forget data, leading to a better trade-off across 2 datasets and 3 unlearning methods.
🧵👇
UPCORE selects a coreset of forget data, leading to a better trade-off across 2 datasets and 3 unlearning methods.
🧵👇
Comments
Yet, standard unlearning can degrade unrelated knowledge, making it unreliable.
Effective unlearning is key for:
📜 Compliance (GDPR, CCPA)
🔐 Privacy & security
⚖️ Ethical AI development
Points contributing to high variance cause more collateral damage when unlearned.
By pruning these outliers, UPCORE reduces unintended forgetting while ensuring effective deletion.
✅ Minimizes unintended degradation
✅ Preserves model utility
✅ Compatible with multiple unlearning methods
Thus, UPCORE reduces negative collateral effects while maintaining effective deletion.
📉 Gradient Ascent
🚫 Refusal
🔄 Negative Preference Optimization (NPO)
We measure:
✔️ Deletion effectiveness – How well the target is removed
✔️ Unintended degradation – Impact on other abilities
✔️ Positive transfer – How well unlearning generalizes
This provides a complete picture of the trade-off between forgetting and knowledge retention over the unlearning trajectory.