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

    Neuronpedia

    INDEX
    Explanations

    (

    np_max-act · gemini-2.0-flash

    This neuron fires on rare or specialized tokens—especially acronyms, proper names (e.g. movie or product titles), and hyphenated or otherwise unusual word‐pieces—i.e. infrequent, domain-specific chunks.

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

    information related to executive compensation and corporate job cuts.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    pray
    -0.06
     사이
    -0.06
    electric
    -0.06
     Besch
    -0.06
     stamped
    -0.06
    rint
    -0.06
     FileAccess
    -0.06
     bande
    -0.06
     Mixer
    -0.06
    HAVE
    -0.06
    POSITIVE LOGITS
     contradictions
    0.06
    .concatenate
    0.06
    ]).↵
    0.06
     nonatomic
    0.06
     النس
    0.06
     Rihanna
    0.06
    Pre
    0.06
    .SuppressLint
    0.06
    .constant
    0.06
    αλ
    0.06
    Act Density 0.292%

    No Known Activations