INDEX
    Explanations

    modelreply if what I say

    structured, technical formatting tokens in text—especially digits (years/section codes), all‑caps acronyms/headers, hyphenated codes, and underscore-delimited identifiers—rather than ordinary prose

    New Auto-Interp
    Negative Logits
     substituents
    0.41
     cufflinks
    0.35
     dimers
    0.35
     substituent
    0.35
     welds
    0.34
     linkage
    0.34
     washers
    0.34
     gusts
    0.33
     craftsmen
    0.33
     cellar
    0.33
    POSITIVE LOGITS
    What
    0.40
    Imagine
    0.34
    Я
    0.34
    1
    0.33
    The
    0.32
    <h1>
    0.31
    #
    0.31
    Кто
    0.31
    Research
    0.31
     What
    0.31
    Act Density 0.741%

    No Known Activations