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

    Neuronpedia

    INDEX
    Explanations

    Code/Documentation

    np_max-act · gemini-2.0-flash

    Spots when the assistant is asking numbered questions or otherwise producing structured, turn-based question prompts (e.g., the "Question 1:", "1.", "Is", and similar Q/A dialog tokens).

    oai_token-act-pair · gpt-5-miniTriggered by @yooniel31
    New Auto-Interp
    Negative Logits
     baby
    -0.07
    -begin
    -0.06
     kids
    -0.06
    μένη
    -0.06
    orne
    -0.06
    jack
    -0.06
     chord
    -0.06
     最
    -0.06
    quant
    -0.06
    .Horizontal
    -0.06
    POSITIVE LOGITS
    .weight
    0.07
    ']]
    0.07
    LIKELY
    0.06
     invariably
    0.06
    :g
    0.06
    _CRYPTO
    0.06
    .updated
    0.06
    .Xna
    0.06
    (parcel
    0.06
     ارزی
    0.06
    Act Density 0.023%

    No Known Activations