training techniqueIntroduced by Prompt tuning · 2021
Soft prompt tuning
Learn only input-level prompt embeddings; it rivals full fine-tuning at scale.
Drafted by AI · not yet reviewed
How this idea evolved
Suggested from citations. No curator has recorded what this idea builds on yet. These ideas from the same theme come from papers that Prompt tuning cites, directly or one step removed. Citation is a fact; the connection between the ideas is not verified.
Optimise a few continuous 'virtual tokens' prepended at every layer of a frozen LM.
Insert small trainable layers into a frozen network, one set per task.