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

    Neuronpedia

    INDEX
    Explanations

    filename or path location

    np_acts-logits-general · gemini-2.5-flash-lite

    This neuron activates on programming keywords, type identifiers, and technical terms in source code, particularly those that define or reference code structures, classes, functions, and data types.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @hkale7

    Programming identifiers, keywords, structural elements, and syntax components within code and technical documentation contexts.

    eleuther_acts_top20 · claude-4-5-sonnetTriggered by @hkale7

    programming code

    np_max-act · claude-4-5-sonnetTriggered by @hkale7

    code

    np_max-act-logits · claude-4-5-sonnetTriggered by @hkale7
    New Auto-Interp
    Negative Logits
    ਖ਼
    -1.91
    ਫ਼
    -1.86
    ਗ਼
    -1.76
    ţei
    -1.55
    立
    -1.37
    來
    -1.34
    ড়
    -1.33
    數
    -1.32
     ţ
    -1.30
    --
    -1.23
    POSITIVE LOGITS
    加拿
    1.38
    â
    1.35
     â
    1.29
    美国的
    1.23
     perdón
    1.23
     häufigsten
    1.23
    [];
    1.23
    שְׁ
    1.23
     puțin
    1.20
    вные
    1.17
    Act Density 3.213%

    No Known Activations