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

    Neuronpedia

    INDEX
    Explanations

    A

    np_max-act · gemini-2.0-flash

    The neuron fires on titlecased tokens—words that are part of headings or titles (e.g. names of works, section headings, proper‐noun headings).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    _comm
    -0.07
    ijn
    -0.07
    ottage
    -0.06
    Opt
    -0.06
     Approx
    -0.06
     Alb
    -0.06
     nowadays
    -0.06
     timestamps
    -0.06
    .Bl
    -0.06
     möchte
    -0.06
    POSITIVE LOGITS
     postId
    0.07
    							↵
    0.06
     bubbles
    0.06
    (dummy
    0.06
     SNAP
    0.06
     Supporters
    0.06
     ↵    ↵
    0.06
     OnCollision
    0.06
    athroom
    0.06
    rank
    0.06
    Act Density 0.054%

    No Known Activations