KRLabsOrg/lettucedect-v2-taxonomy-head

Update README.md

fix: restore jhu-clsp/mmBERT-base tokenizer_config (transformers 4.x compatibility; v5 save artifacts removed)

fix: tokenizer_class TokenizersBackend -> PreTrainedTokenizerFast (transformers 4.x compatibility)

Add LettuceCode mascot + tagline

Card: training method + data

lettucedect-v2-taxonomy-head: span typing head for the encoder cascade

initial commit