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

    Neuronpedia

    INDEX
    Explanations

    late 2023 / early 2024

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

    references to specific calendar years and dated time markers in text.

    oai_token-act-pair · gpt-5Triggered by @yooniel31
    New Auto-Interp
    Negative Logits
     sometimes
    0.62
    НЫ
    0.52
     uses
    0.52
     any
    0.51
    0.51
     nėra
    0.48
     different
    0.47
              
    0.46
     always
    0.46
     likes
    0.46
    POSITIVE LOGITS
    年初
    0.68
     році
    0.63
     onwards
    0.62
     ਹੀ
    0.62
    年底
    0.61
    umumkan
    0.61
     anunció
    0.61
     aveva
    0.60
    下旬
    0.59
     commenced
    0.58
    Act Density 0.020%

    No Known Activations