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

    Neuronpedia

    INDEX
    Explanations

    speech

    np_max-act · gemini-2.0-flash

    technical concepts and services related to middleware and web hosting.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8

    The neuron fires on prompt phrasing that asks the model to write a “speech” (and related framing words like “as,” “of,” “by,” or “spoken”)—in other words, it detects instructions to generate a speech.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
     fold
    -0.06
    кових
    -0.06
     століття
    -0.06
     vhodné
    -0.06
     practices
    -0.06
    Akt
    -0.06
     trouble
    -0.06
     工
    -0.06
    _value
    -0.06
     readers
    -0.06
    POSITIVE LOGITS
     speeches
    0.07
    othermal
    0.07
     <",
    0.07
     Speech
    0.07
     speech
    0.07
     Perm
    0.07
    ToLower
    0.07
    ُو
    0.06
     par
    0.06
     Necklace
    0.06
    Act Density 0.011%

    No Known Activations