“Be like a Goldfish, Don’t Memorize! Mitigating Memorization in Generative LLMs”, Abhimanyu Hans, Yuxin Wen, Neel Jain, John Kirchenbauer, Hamid Kazemi, Prajwal Singhania, Siddharth Singh, Gowthami Somepalli, Jonas Geiping, Abhinav Bhatele, Tom Goldstein2024-06-14 (, ; similar)⁠:

Large language models can memorize and repeat their training data, causing privacy and copyright risks.

To mitigate memorization, we introduce a subtle modification to the next-token training objective that we call the goldfish loss. During training, a randomly sampled subset of tokens are excluded from the loss computation. These dropped tokens are not memorized by the model, which prevents verbatim reproduction of a complete chain of tokens from the training set.

We run extensive experiments training billion-scale Llama-2 models, both pre-trained and trained from scratch, and demonstrate reductions in extractable memorization with little to no impact on downstream benchmarks.