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

    Neuronpedia

    INDEX
    Explanations

    self-referential text

    np_max-act · gemini-2.0-flash

    tokens that appear in assistant self-descriptions (mentions of "AI"/"language model", training/knowledge cutoff, updates, system time and related limitations).

    oai_token-act-pair · gpt-5-miniTriggered by @yooniel31

    AI

    np_max-act-logits · gpt-5-miniTriggered by @yooniel31

    Explanation could not be parsed.

    eleuther_acts_top20 · gpt-5-miniTriggered by @yooniel31
    New Auto-Interp
    Negative Logits
    kate
    -0.06
    lapping
    -0.06
    -fin
    -0.06
     anz
    -0.06
    kal
    -0.06
    iors
    -0.06
    شود
    -0.06
    ploy
    -0.06
    (rawValue
    -0.06
    -know
    -0.06
    POSITIVE LOGITS
     Digital
    0.08
     alliances
    0.07
     accelerometer
    0.06
     Autodesk
    0.06
    DlgItem
    0.06
     firstly
    0.06
    +'</
    0.06
     Warranty
    0.06
     flex
    0.06
     Arth
    0.06
    Act Density 0.035%

    No Known Activations