framework
active
framework:backpack-language-models

Backpack Language Models

LM architecture with sense vectors showing multiplication effects, illustrating custom intervention in pyvene

Neighborhood — ranked by edge-count

Papers (1)

paper

Thinkers (1)

thinker
  • John Hewitt
    studies
    Construction manager and chief engineer for the West Dean Visitor's Centre, demonstrating program budgeting success.

Concepts (1)

concept
  • Sense Vectors
    implements
    Vectors acquired during pretraining in Backpack LMs that have a multiplication effect on model generation

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

  • Language Modelsconcept0.845
    Primary substrate for manifold steering experiments; demonstrates method on reasoning and in-context tasks.
  • Language Modelconcept0.835
    Primary test domain for manifold steering, including reasoning and ICL tasks
  • modelconcept0.758
    A representation that captures relevant aspects of a system; according to the theorem, the regulator must embody this.
  • Bias in language modelsconcept0.753
    Features related to gender, racial, ethnic biases, slurs, and hate speech.
  • Role-play model of large language modelsframework0.727
    Framework describing LLMs as role-play engines, introduced in Shanahan, McDonell, Reynolds 2023.
  • Large Language Models (LLMs)concept0.716
    Transformer-based models like GPT-4, LaMDA, PaLM; assessed for GWT indicators.
  • Manual model editingconcept0.715
    Ability to surgically alter model behavior through direct parameter changes rather than activation interventions.
  • Toy Modelsconcept0.712