INDEX
    Explanations

    Dates and times

    This neuron fires on the ordinal suffixes (rd, th, nd) and punctuation used in date stamps, i.e. it detects the tokens marking day numbers in dates.

    New Auto-Interp
    Negative Logits
     Hale
    -0.07
     toda
    -0.07
     FP
    -0.06
    ा�
    -0.06
    .Batch
    -0.06
     Servers
    -0.06
    uan
    -0.06
    clazz
    -0.06
     offsets
    -0.06
     CFR
    -0.06
    POSITIVE LOGITS
    0.07
    เคราะห
    0.07
    начала
    0.06
     prescriptions
    0.06
     Trie
    0.06
     čís
    0.06
     mutating
    0.06
     inds
    0.06
     UNITED
    0.06
    YPRE
    0.06
    Act Density 0.004%

    No Known Activations