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

    Neuronpedia

    INDEX
    Explanations

    presenting options

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

    phrases that indicate conversational help requests or structured, step-by-step/option-based responses in a chat-style exchange.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
    ーター
    0.39
    ISTICS
    0.38
    㚅
    0.36
     umfang
    0.36
     vollständig
    0.35
    рактери
    0.34
     devons
    0.34
     egensk
    0.34
     habilidades
    0.34
     uitgebre
    0.34
    POSITIVE LOGITS
     you
    0.48
     Yes
    0.48
     Doesn
    0.48
    ใช่
    0.48
     Yeah
    0.47
     yeah
    0.46
     Seems
    0.46
     honestly
    0.43
     your
    0.43
     admittedly
    0.43
    Act Density 0.178%

    No Known Activations