dataset
active
dataset:claude-haiku-4-5

Claude Haiku 4.5

Target model for NLA training and evaluation; shows similar FVE curves to Haiku 3.5.

Neighborhood — ranked by edge-count

Papers (1)

paper

Methods (1)

method
  • Natural Language Autoencoders (NLAs)
    aboutimplements
    Core unsupervised method for generating natural language explanations of LLM activations through a verbalizer-reconstructor pair trained with RL.

Institutes (1)

institute
  • Anthropic
    associated_with
    Lab behind Claude models and Constitutional AI training approach; represents highest baseline scores and lowest prompt lift.