This is a fine-tuned version of microsoft/deberta-v3-small for predicting Patent CPC codes.

Dataset consists of titles and abstracts sampled from granted patent applications: https://www.kaggle.com/datasets/grimmace/sampled-patent-titles