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

    Neuronpedia

    INDEX
    Explanations

    attention mechanism function

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     signos
    0.43
    له
    0.41
     نفسها
    0.41
     Stevens
    0.40
    Within
    0.40
    他們
    0.40
     nucleon
    0.40
    They
    0.39
    ب
    0.39
    Dow
    0.38
    POSITIVE LOGITS
     нужен
    0.47
     треба
    0.43
     umožňuje
    0.43
    浸
    0.43
     distinguishes
    0.39
    preprocess
    0.39
    最重要的
    0.39
     prenez
    0.39
     crucial
    0.39
     phải
    0.39
    Act Density 0.005%

    No Known Activations