Figure 8
![An illustration of (a) LM pre-training, (b) standard fine-tuning, and (c) discrete prompt-based fine-tuning (or prompt-tuning) [43]. Refer to the image caption for details.](https://emer.silverchair-cdn.com/emer/content_public/journal/atsip/13/2/10.1561_116.00000010/3/116.00000010008.png?Expires=1782488868&Signature=x4fpwmMCJJk0DfC5JhSfj1NgvGCeWWZ0VJcsuFx36S9Ywc8sgxkjwptaFqCe8YNxKnPOUIjEuPgqIdrWjQIPRmLC24541G8BvnftrTc-T9wyIPMEcRGoVuJmZNoQHNIbOGkVDvfzc-koj-ovbYO3UM2o4b3wi-Rq8vxqBqs0SjjIxRpPnn165je9VaiAvNMyc-F-U5VGn1~tMoWsj4k949NDjIE6Ly-pHfzA~aks~Y9HjA3P2y4XC9G65rDN0Mmp2E8DnOBQIjvh~ZVn2ejHSF7~2YJD7XoJ4HiHVnewjI97YZ3cPnQqgySxmNb5bdZfkLFcos~AJ0FDavNDCR0jbA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
An illustration of (a) LM pre-training, (b) standard fine-tuning, and (c) discrete prompt-based fine-tuning (or prompt-tuning) [43].
An illustration of (a) LM pre-training, (b) standard fine-tuning, and (c) discrete prompt-based fine-tuning (or prompt-tuning) [43].