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

    Neuronpedia

    INDEX
    Explanations

    Controversial social/political discussions

    np_max-act · gemini-2.0-flash

    The neuron activates on Spanish text segments (Spanish words and punctuation).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    finds tokens that indicate abusive or harmful content, especially hate speech, racist/discriminatory claims, or calls for violence/self-harm.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0

    content involving hate speech or slurs and sensitive discussions about race and discrimination, often alongside other safety-flag topics like violence or self-harm.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
    rnd
    -0.07
    ивать
    -0.07
    raud
    -0.07
    factor
    -0.07
     separ
    -0.06
     technically
    -0.06
    λογή
    -0.06
    ifica
    -0.06
    Email
    -0.06
     impeachment
    -0.06
    POSITIVE LOGITS
    -submit
    0.07
    REDIT
    0.06
    ař
    0.06
    朋
    0.06
     descri
    0.06
    Country
    0.06
     Az
    0.06
     Slot
    0.06
     slots
    0.06
     commem
    0.06
    Act Density 0.103%

    No Known Activations