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

    Neuronpedia

    INDEX
    Explanations

    fixing or preventing problems

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
     Engines
    0.34
    强化
    0.31
     Alternative
    0.31
    urized
    0.31
    Alternative
    0.31
     couro
    0.30
    尝试
    0.29
    统一
    0.29
     γν
    0.29
     чувстви
    0.29
    POSITIVE LOGITS
     errors
    1.08
    errors
    0.95
     mistakes
    0.88
     problems
    0.85
     mishaps
    0.85
     inconsistencies
    0.85
     errores
    0.84
     flaws
    0.84
     inaccuracies
    0.81
     erreurs
    0.80
    Act Density 0.087%

    No Known Activations