Truly Sparse Neural Networks at Scale
This work is significant for researchers and practitioners in deep learning who are looking to improve the efficiency and scalability of neural network training, especially for very large models.
This paper tackles the problem of inefficient sparse neural network training by introducing a parallel training algorithm, a specialized activation function, and a neuron importance metric. They successfully trained the largest neural network to date, achieving state-of-the-art performance.
Recently, sparse training methods have started to be established as a de facto approach for training and inference efficiency in artificial neural networks. Yet, this efficiency is just in theory. In practice, everyone uses a binary mask to simulate sparsity since the typical deep learning software and hardware are optimized for dense matrix operations. In this paper, we take an orthogonal approach, and we show that we can train truly sparse neural networks to harvest their full potential. To achieve this goal, we introduce three novel contributions, specially designed for sparse neural networks: (1) a parallel training algorithm and its corresponding sparse implementation from scratch, (2) an activation function with non-trainable parameters to favour the gradient flow, and (3) a hidden neurons importance metric to eliminate redundancies. All in one, we are able to break the record and to train the largest neural network ever trained in terms of representational power -- reaching the bat brain size. The results show that our approach has state-of-the-art performance while opening the path for an environmentally friendly artificial intelligence era.