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

    Neuronpedia

    INDEX
    Explanations

    numbers

    np_max-act · gemini-2.0-flash

    The neuron activates on numeric tokens—particularly those containing decimal points (e.g. “0.615234375,” “2.154296875,” etc.). It’s essentially spotting floating-point numbers.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    _dest
    -0.07
    END
    -0.07
     ending
    -0.07
    istrator
    -0.07
     insert
    -0.06
     structures
    -0.06
    	request
    -0.06
     frec
    -0.06
    Internal
    -0.06
    Regression
    -0.06
    POSITIVE LOGITS
    nat
    0.08
    .MessageBox
    0.06
     outweigh
    0.06
     söyledi
    0.06
     �
    0.06
    urv
    0.06
    '),↵↵
    0.06
    .^
    0.06
    dropout
    0.06
    .eu
    0.06
    Act Density 0.108%

    No Known Activations