This dataset is a combined, enriched version of the self_dialog and woz_dialog splits from the Taskmaster-1 dataset. It consists of multi-turn, human-human and human-simulated conversations with systematic enhancements for machine learning workflows—especially dialog modeling, generation, and fine-grained evaluation.
All conversations are structured in a JSON format with consistent schema and include added semantic… See the full description on the dataset page: https://huggingface.co/datasets/GenAIDevTOProd/taskmaster_v1_enriched.
