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

    Neuronpedia

    INDEX
    Explanations

    word endings and subsequent punctuation

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     was
    0.49
    defender
    0.43
     glistening
    0.42
     Се
    0.42
    ли
    0.42
     아
    0.42
     in
    0.41
     glazed
    0.41
    ила
    0.41
     เรีย
    0.41
    POSITIVE LOGITS
    ers
    0.53
    ↵
    0.51
    able
    0.51
    B
    0.50
    ings
    0.49
    र
    0.47
    ments
    0.46
    INGS
    0.45
    ін
    0.45
    者を
    0.45
    Act Density 0.144%

    No Known Activations