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

    Neuronpedia

    INDEX
    Explanations

    crime/law enforcement

    np_max-act · gemini-2.0-flash

    Neuron 1: looks for actions or phrases in present participle form describing ongoing, dynamic activity. Neuron 2: looks for terms related to medical conditions or health risks. Neuron 3: looks for phrases about a sense of community and collective togetherness. Neuron 4: looks for content that promotes engagement and activism through social media and citizen journalism.

    oai_token-act-pair · gpt-5-nanoTriggered by @tatsatx

    spots proper nouns, names, organizations and header-like tokens (i.e., named entities and section/headline text).

    oai_token-act-pair · gpt-5-miniTriggered by @tatsatx
    New Auto-Interp
    Negative Logits
     prag
    -0.07
    上述
    -0.07
     reproduced
    -0.07
    ，谢谢
    -0.07
     :)↵↵
    -0.07
    谨
    -0.07
    总结
    -0.07
     bovenstaande
    -0.07
     rhetorical
    -0.06
     extrap
    -0.06
    POSITIVE LOGITS
    &apos
    0.11
    &lt
    0.10
    &#
    0.10
     &#
    0.10
    &nbsp
    0.09
    <br
    0.08
    &amp
    0.08
    /foo
    0.08
    &eacute
    0.08
    配送
    0.08
    Act Density 46.865%

    No Known Activations