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

    Neuronpedia

    INDEX
    Explanations

    The neuron detects programming source code (code blocks and code tokens/structures) within the document.

    oai_token-act-pair · gpt-5-miniTriggered by @fornasiereraffaello
    New Auto-Interp
    Negative Logits
    Diffusion
    0.68
    Lys
    0.63
    Necess
    0.63
    Charting
    0.63
    cwnd
    0.62
    Ś
    0.62
    Hitpoint
    0.62
    Dried
    0.61
    Bias
    0.61
    Mits
    0.60
    POSITIVE LOGITS
     create
    1.08
     generate
    1.07
     get
    1.03
     initialize
    1.00
     calculate
    0.94
     validate
    0.92
     isValid
    0.91
     addNew
    0.91
     init
    0.90
     handle
    0.89
    Act Density 0.319%

    No Known Activations