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

    Neuronpedia

    INDEX
    Explanations

    question answering

    np_max-act · gemini-2.0-flash

    This neuron detects parameter labels or section headings in the user’s instructions, specifically words followed by a colon (e.g., “Ort:”).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    requirements
    -0.09
     Sofia
    -0.07
     مراج
    -0.07
     headers
    -0.07
     взаєм
    -0.07
     almaktadır
    -0.07
    ół
    -0.06
     мереж
    -0.06
     madde
    -0.06
     casi
    -0.06
    POSITIVE LOGITS
    λλη
    0.07
     yak
    0.07
    apeutic
    0.07
    بل
    0.06
    Risk
    0.06
    เพลง
    0.06
     convert
    0.06
     habit
    0.06
    berra
    0.06
    prom
    0.06
    Act Density 0.008%

    No Known Activations