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

    Neuronpedia

    INDEX
    Explanations

    generate, draw, slightly

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

    identifiers and tokens from code snippets, especially snake_case function/variable names with underscores and related code-format elements.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Negative Logits
    ERON
    0.38
    Zed
    0.37
    летие
    0.34
    BSD
    0.33
    MBOLS
    0.33
    ات
    0.33
    ATIV
    0.33
    вших
    0.32
    ATIONS
    0.32
    GSM
    0.32
    POSITIVE LOGITS
             
    0.33
     थोड़ा
    0.33
    0.33
     slightly
    0.32
     좀
    0.31
     Democrat
    0.30
     tn
    0.30
     कोशिश
    0.30
    _
    0.30
     Slightly
    0.29
    Act Density 0.552%

    No Known Activations