An Integrated Framework for Controllable Text Generation

Explore an integrated framework for controllable text generation in this seminar series talk by Wei Xu, an assistant professor at Georgia Institute of Technology. Delve into the challenges of natural language generation and learn about innovative solutions, including high-quality data construction, controllable neural generation models, interactive annotation interfaces, and redesigned evaluation methodologies. Discover the LENS learnable evaluation metric, which outperforms existing metrics in correlating with human judgments and serves as an effective learning objective for creating state-of-the-art neural generation models. Gain insights into text simplification, scientific writing, and the application of deep learning techniques in natural language processing.