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

    Neuronpedia

    INDEX
    Explanations

    .from

    np_max-act · gemini-2.0-flash

    This neuron detects code tokens involved in loading pretrained models, in particular the “.from_pretrained” method calls.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Negative Logits
    ikk
    -0.07
     Festival
    -0.07
    _print
    -0.07
     هند
    -0.07
    .att
    -0.07
    �
    -0.06
     extraction
    -0.06
     Strateg
    -0.06
     visitors
    -0.06
     bryster
    -0.06
    POSITIVE LOGITS
    _than
    0.06
    Cancelable
    0.06
     Animals
    0.06
     jsonResponse
    0.06
     Common
    0.06
    (entry
    0.06
     }
↵
↵
↵
↵
    0.06
    TARGET
    0.06
    0.06
     conj
    0.06
    Act Density 0.002%

    No Known Activations