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

    Neuronpedia

    INDEX
    Explanations

    pseudo suffixes

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

    The neuron detects salient, content-bearing words (strongly descriptive or emphatic nouns/adjectives/keywords) that carry the main topic or emphasis in a passage.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Negative Logits
     financière
    0.44
     stately
    0.44
     supremely
    0.43
     overpowering
    0.41
     vitally
    0.41
     professe
    0.40
     grandeur
    0.40
     felicidad
    0.40
     almighty
    0.40
     majest
    0.39
    POSITIVE LOGITS
    Pseudo
    0.67
    pseudo
    0.66
     pseudo
    0.60
     Pseudo
    0.58
    (?)
    0.57
     indirectly
    0.57
    ish
    0.56
    albeit
    0.56
    esque
    0.54
     indire
    0.54
    Act Density 0.953%

    No Known Activations