I'm seeking an experienced machine learning engineer to help debug a beta - variational autoencoder (VAE). The VAE is encountering training errors, specifically in the training loop, and the model is not converging. I need help creating code for a subset of data and to implement KL annealing. Requirements: - Expertise in neural networks and VAEs. - Strong understanding of training loops in deep learning. - Experience with debugging training issues. - Proficiency in Python and frameworks like TensorFlow or PyTorch.