Classifiers are Better Experts for Controllable Text Generation

This paper proposes a simple method for controllable text generation based on weighting logits produced, namely CAIF sampling. Using an arbitrary third-party text classifier, we adjust a small part of a language model's logits and guide text generation towards or away from classifier prediction. We show that the proposed method significantly outperforms recent PPLM, GeDi, and DExperts on PPL and sentiment accuracy based on the external classifier of generated texts. A the same time, it is also easier to implement and tune, and has significantly fewer restrictions and requirements.

Variational Autoencoders for Studying the Manifold of Precoding Matrices with High Spectral Efficiency

In multiple-input multiple-output (MIMO) wireless communications systems, neural networks have been employed for channel decoding, detection, channel estimation, and resource management. In this paper, we look at how to use a variational autoencoder to find a precoding matrix with a high Spectral Efficiency (SE). To collect optimal precoding matrices, an optimization approach is used. Our objective is to create a less time-consuming algorithm with minimum quality degradation. To build precoding matrices, we employed two forms of variational autoencoders: conventional variational autoencoders (VAE) and conditional variational autoencoders (CVAE). Both methods may be used to study a wide range of optimal precoding matrices. To the best of our knowledge, the development of precoding matrices for the spectral efficiency objective function (SE) utilising VAE and CVAE methods is being published for the first time.