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

    Neuronpedia

    INDEX
    Explanations

    emergence and persistence of states

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    Patch
    0.43
     celebratory
    0.41
     Sequences
    0.40
    roller
    0.39
     walking
    0.38
     Targeted
    0.38
    zh
    0.38
     pretend
    0.38
    例如
    0.37
     Walking
    0.37
    POSITIVE LOGITS
     arises
    0.75
     arose
    0.73
     arisen
    0.62
     arise
    0.58
     persists
    0.57
     lingers
    0.54
     возникает
    0.53
     outweighs
    0.52
     culminated
    0.52
     lessened
    0.51
    Act Density 0.045%

    No Known Activations