Question
What is model memorization?
By SourceX Editorial · Updated
Short answer
Memorization is when a model stores specific training examples closely enough to reproduce them. It is more likely for text that appears many times or is unusual. It is why sensitive details should be removed before data is used for training.
Part of: How licensing works
Explanation#
Labs use deduplication and filtering to reduce it.
Contracts can prohibit attempts to extract training data.
Rights and privacy#
- Confirm customer, vendor and employee agreements allow the use
- Remove or de-identify personal details before anything leaves the company
- Approve the final dataset and permitted uses in writing
How licensing works#
A business describes its data, buyer demand is assessed, scope and permitted use are proposed, the data is prepared and de-identified, an agreement is signed, and the approved dataset is delivered. See data licensing for detail.
Related questions
Sources
General information, not legal advice. Editorial policy.
Related resources
- SolutionWhat is AI evaluation data?
- SolutionHow AI developers source data
- IndustryBPO & contact centers data
- IndustryRecruiting & staffing data
- DataCode review records
- DataCompany email archives
- GlossaryRevenue share
- QuestionShould companies sell or license their data?
- QuestionCan CRM data be licensed?
- InsightIs it safe to license company data for AI training?
- InsightHow does AI training data licensing work?
- InsightHow is company data anonymized before AI training?
See if your company qualifies
A short company assessment. No data uploads are needed.