3 ms·
Can an already trained LLM be made (fine-tuned?) to forget a specific document, by running gradient descent in reverse?
by andai 1y ago
Can an already trained LLM be made (fine-tuned?) to forget a specific document, by running gradient descent in reverse?