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

    Neuronpedia

    INDEX
    Explanations

    predicting masked words

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    濛
    0.44
     contradictory
    0.44
     firmly
    0.43
    ந
    0.43
    рьох
    0.43
     சின்ன
    0.42
     horrified
    0.41
     shocking
    0.41
    มิ
    0.41
     keď
    0.41
    POSITIVE LOGITS
     nucl
    0.47
    spiration
    0.47
    it
    0.46
     Drill
    0.44
    er
    0.43
     alpin
    0.43
     بطور
    0.42
     almac
    0.41
     almacen
    0.41
     livet
    0.41
    Act Density 0.004%

    No Known Activations