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

    Neuronpedia

    INDEX
    Explanations

    Gemma team at Google DeepMind

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

    mentions of the model name "Gemma" or tokens referring to the assistant's identity.

    oai_token-act-pair · gpt-5-miniTriggered by @yooniel31
    New Auto-Interp
    Negative Logits
    p
    2.58
    a
    2.31
    of
    2.02
    de
    1.77
    b
    1.73
    an
    1.70
    c
    1.68
    g
    1.60
    y
    1.41
    sp
    1.40
    POSITIVE LOGITS
    م
    1.19
    ер
    1.08
    м
    1.06
     fertil
    1.05
     großen
    1.04
    নি
    1.03
    (),
    1.02
     různých
    1.02
     sơn
    1.01
    ુર
    0.98
    Act Density 0.620%

    No Known Activations