Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    elements related to personalization and creative expression in writing

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    Follows "a" or modifies an adjective

    np_acts-logits-general · gemini-2.0-flash

    a twist or spin

    np_acts-logits-general · gemini-2.5-flash-lite

    The marked phrases represent adjectives or adjectival phrases used to describe modifications, enhancements, or distinctive qualities added to nouns. These phrases typically convey customization, personalization, or refinement—such as "with a twist," "a fresh spin," "added sophistication," "an extra layer," or "a welcome shot." The pattern demonstrates how descriptive language emphasizes the value-addition or improvement of the preceding noun, creating emphasis on uniqueness, style, or enhancement.

    eleuther_acts_top20 · claude-4-5-haikuTriggered by @kparkhamchuk
    New Auto-Interp
    Negative Logits
    SharedDtor
    -1.06
    IntoConstraints
    -0.77
    ThroughAttribute
    -0.77
    RenderAtEndOf
    -0.72
     kasarigan
    -0.71
    CodeAttribute
    -0.69
    uesia
    -0.69
    WireFormatLite
    -0.68
    endpush
    -0.67
    TableAdapter
    -0.66
    POSITIVE LOGITS
     touch
    1.97
     touches
    1.76
     twist
    1.63
     touche
    1.47
    touch
    1.46
     TOUCH
    1.40
     Touch
    1.36
    Touch
    1.33
     touched
    1.23
     twists
    1.22
    Act Density 0.192%

    No Known Activations