An NLP annotation tool that can run directly in the browser. It allows for annotation of multiple tasks on a single dataset and supports four task-types: sequence labeling, span labeling, text classification and seq2seq.
A Bavarian Universal Dependency treebank with 15k annotated tokens from all Bavarian dialect areas and multiple text genres (wiki, fiction, grammar examples, social, non-fiction).
NaBaLiSID (Natural Lithuanian and Bavarian Slot and Intent Detection) provides new slot and intent detection evaluation datasets for Bavarian and Lithuanian, combining translations of xSID and MASSIVE with more natural, non-translated utterances.
An overview of 50+ datasets with human label variation (multiple, un-aggregated annotations per instance) in Natural Language Processing and Computer Vision.