Comment by 29athrowaway

No, we are not. Pretrained models do not learn from their prompts. Their state is volatile.