A single color token becomes exact integer indicators.

highlighted = computed this step

Read row A

Row A has color red. In the pinned column order, the red indicator is 1 and the other indicators are 0.

Ared=(1,0,0,0)A_{\text{red}}=(1,0,0,0)
Categorical Features ExactlyExact one-hot columns from a pinned category list.one-hot columns from pinned category orderrowcolorredbluegreenotherAred1000Bblue0100Cgreen0010category order: red, blue, green; other reservedpreprocessing rewrites category tokens into indicator columnsunknown category uses the pinned other-column policyNOT training; NOT learning; NOT generalizationNOT accuracy; NOT calibration; NOT a guarantee of model qualityNOT probability truth

Exactly one known column is on

For a known category row, exactly one known-category column gets 1. The other column stays 0.

1+0+0+0=11 + 0 + 0 + 0 = 1
Categorical Features ExactlyExact one-hot columns from a pinned category list.one-hot columns from pinned category orderrowcolorredbluegreenotherAred1000Bblue0100Cgreen0010category order: red, blue, green; other reservedpreprocessing rewrites category tokens into indicator columnsunknown category uses the pinned other-column policyNOT training; NOT learning; NOT generalizationNOT accuracy; NOT calibration; NOT a guarantee of model qualityNOT probability truth

The token is not interpreted

The word red is not turned into meaning. It is matched to the pinned red column.

red tokenred indicator\text{red token}\rightarrow\text{red indicator}
Categorical Features ExactlyExact one-hot columns from a pinned category list.one-hot columns from pinned category orderrowcolorredbluegreenotherAred1000Bblue0100Cgreen0010category order: red, blue, green; other reservedpreprocessing rewrites category tokens into indicator columnsunknown category uses the pinned other-column policyNOT training; NOT learning; NOT generalizationNOT accuracy; NOT calibration; NOT a guarantee of model qualityNOT probability truth