Card: link the zero-shot leaderboard

Card: two more unseen-label benchmarks (MTOP 6 languages, CLINC150)

Card: base v1.3 numbers

v1.3: native-language labels (+0.043 MASSIVE / +0.045 SIB-200 with labels in the text's language)

Card: native-language label evaluation

Card: link training dataset

Card: base column now v1.2

Add v1.1 data generator code

Card: add large (568M) column

Card: widget examples

v1.1: generic-taxonomy training data (see card: v1.0 -> v1.1)

Add zero-shot top-label agreement check for int8 ONNX

Card: label-wording limitation

Release multilingual zero-shot classifier v1.0 (small)

initial commit