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

    Neuronpedia

    INDEX
    Explanations

    programmed refusal

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     basically
    0.48
    basically
    0.45
     basics
    0.43
    uo
    0.40
    Basically
    0.38
     Basically
    0.38
    IGet
    0.38
     migr
    0.37
     anus
    0.37
    也沒有
    0.37
    POSITIVE LOGITS
     sorry
    0.66
     Sorry
    0.65
    Sorry
    0.60
    sorry
    0.57
     सॉरी
    0.56
     regret
    0.53
     regretted
    0.53
     regrett
    0.47
     apology
    0.46
     сожалению
    0.43
    Act Density 0.004%

    No Known Activations