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

    Neuronpedia

    INDEX
    Explanations

    issues continuing

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     remnant
    0.46
     overestimate
    0.42
     residual
    0.42
     mismanagement
    0.41
     farewell
    0.40
    >≤</
    0.39
     compensatory
    0.38
     durchaus
    0.38
    なくなる
    0.38
     actually
    0.37
    POSITIVE LOGITS
     STILL
    1.30
     Still
    1.19
     still
    1.15
    Still
    1.13
    still
    1.13
    还是
    1.03
    依然
    0.99
     vẫn
    0.96
    還是
    0.96
    仍然
    0.95
    Act Density 0.005%

    No Known Activations