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

    Neuronpedia

    INDEX
    Explanations

    suspicion and corroboration

    np_acts-logits-general · gemini-2.5-flash-lite

    phrases that introduce illustrative examples or clarifying parenthetical asides, often via quoted sample statements or list-style elaboration.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
     będ
    0.40
    Ready
    0.39
    Essentially
    0.38
     ensures
    0.37
    也将
    0.37
    将会
    0.37
     möchte
    0.36
     giúp
    0.36
     થશે
    0.36
    Efficiency
    0.36
    POSITIVE LOGITS
     suspiciously
    0.95
     suspicious
    0.93
     suspicions
    0.78
     plausible
    0.77
     plaus
    0.72
     suspicion
    0.71
     sospe
    0.68
     corrobor
    0.67
     circumstantial
    0.64
     suspected
    0.63
    Act Density 0.268%

    No Known Activations