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

    Neuronpedia

    INDEX
    Explanations

    key performance improvements

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     rc
    1.63
     Method
    1.54
     Mathew
    1.52
    Method
    1.52
     Mathews
    1.51
    Mat
    1.50
     Кре
    1.44
    mat
    1.44
    rat
    1.41
     Rit
    1.40
    POSITIVE LOGITS
     B
    1.38
    B
    1.10
    8
    0.96
    7
    0.93
    星球
    0.92
    比べ
    0.90
     Tauri
    0.89
     benda
    0.86
     Hunde
    0.84
     beban
    0.80
    Act Density 0.041%

    No Known Activations