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

    Neuronpedia

    INDEX
    Explanations

    Code snippets

    np_max-act · gemini-2.0-flash

    The neuron is looking for conversation “role” or section markers (e.g. “Q:”, “A:”, “<|begin_of_text|>”, etc.) that delimit turns or code blocks in the dialogue.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    memcpy
    -0.06
    ailable
    -0.06
    라고
    -0.06
    Responsive
    -0.06
    лений
    -0.06
    因为
    -0.06
     Discipline
    -0.06
    满
    -0.06
    苹果
    -0.06
    :aload
    -0.06
    POSITIVE LOGITS
    (Op
    0.07
     üretim
    0.06
    (limit
    0.06
    (dx
    0.06
     протягом
    0.06
    ічного
    0.06
     ई
    0.06
    ƒ
    0.06
     zk
    0.06
     système
    0.06
    Act Density 0.022%

    No Known Activations