Plug-and-Play Language Model (PPLM)

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Role-play model of large language modelsframework0.766
  Framework describing LLMs as role-play engines, introduced in Shanahan, McDonell, Reynolds 2023.
• Language Modelconcept0.750
  Primary test domain for manifold steering, including reasoning and ICL tasks
• Language Modelsconcept0.739
  Primary substrate for manifold steering experiments; demonstrates method on reasoning and in-context tasks.
• Language models are some of the most remarkable computer programs in existence.quote0.739
  Opening sentence setting the stage for the importance of interpretability.
• Language models are few-shot learners (Brown et al., 2020)concept0.738
  Demonstrated transformers on mathematical understanding and logic; cited to motivate transformer versatility.
• Large Language Models (LLMs)concept0.737
  Transformer-based models like GPT-4, LaMDA, PaLM; assessed for GWT indicators.
• In some sense, this is the simplest language model we profoundly don't understand. And so it makes a natural target for our paper.quote0.734
  Articulates why a one-layer transformer with MLP is the appropriate starting target for mechanistic interpretability
• PP-1: Input modules using predictive codingconcept0.733
  Indicator from PP: use of predictive coding in input modules.