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

    Neuronpedia

    INDEX
    Explanations

    conjunctions and words indicating contrast or alternatives

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    preceding "though" or "but"

    np_acts-logits-general · gemini-2.0-flash

    but or though introducing exceptions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Negative Logits
    imed
    -0.49
    such
    -0.43
     Namely
    -0.43
    ẓ
    -0.43
    arak
    -0.43
    MigrationBuilder
    -0.42
     Such
    -0.42
    事的
    -0.41
     relatifs
    -0.41
     amarillas
    -0.41
    POSITIVE LOGITS
     occasionally
    1.38
    Occasionally
    1.32
     Occasionally
    1.28
     sometimes
    1.17
    occasion
    1.10
     occasional
    1.06
    sometimes
    1.02
     Sometimes
    1.00
    偶尔
    0.99
    たまに
    0.98
    Act Density 0.507%

    No Known Activations