Category classifier training datasets organized by input representation. Each variant keeps the same CSV schema and uses embedding_text as the model input for scripts/hf_train_category.sh.
category/title/ner-v2-cleaned/dataset.filtered.csv - cleaned title only.
category/title-plus-ner/ner-v2-cleaned/dataset.filtered.csv - cleaned title plus legacy NER fields.
category/no-title-plus-ner/ner-v2-cleaned/dataset.filtered.csv - legacy… See the full description on the dataset page:
https://huggingface.co/datasets/buyanov-tech/product-classifier-category-data.
