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

    Neuronpedia

    INDEX
    Explanations

    Frustration and help requests

    np_max-act · gemini-2.0-flash

    This neuron activates on self-referential phrases where the author talks about their own situation or feelings (e.g. “I am new,” “I’m out of ideas,” “help would be appreciated”).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    first-person requests for help or expressions of being stuck/new (pleas for assistance or frustration).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
     formatted
    -0.07
     acre
    -0.07
    Ж
    -0.06
     weaponry
    -0.06
    Commercial
    -0.06
     sple
    -0.06
    طن
    -0.06
    done
    -0.06
     milk
    -0.06
    _music
    -0.06
    POSITIVE LOGITS
    .production
    0.07
     ابن
    0.06
    �
    0.06
     ‪
    0.06
     itemprop
    0.06
    ("***
    0.06
     Registers
    0.06
     savedInstanceState
    0.06
    假
    0.06
     UserDefaults
    0.06
    Act Density 0.071%

    No Known Activations