INDEX
    Explanations

    as if or like comparisons

    New Auto-Interp
    Negative Logits
     relatively
    0.50
     typically
    0.44
     relativamente
    0.43
     likely
    0.42
    relatively
    0.41
    可谓
    0.41
    typically
    0.41
    看似
    0.41
     अपेक्षाकृत
    0.39
     ricor
    0.38
    POSITIVE LOGITS
     будто
    0.58
     underwater
    0.57
     somehow
    0.55
     invisible
    0.51
    不存在
    0.50
     simultaneously
    0.49
     одновременно
    0.48
     동시에
    0.48
    someone
    0.47
     যেন
    0.46
    Act Density 0.101%

    No Known Activations