Sample-based Dynamic Hierarchical Transformer with Layer and Head Flexibility via Contextual Bandit
This addresses the problem of inefficient training and inference in Transformers for AI practitioners, offering a novel dynamic approach rather than incremental compression.
The paper tackles the inflexibility and computational expense of fixed-layer/head Transformers by proposing a sample-based Dynamic Hierarchical Transformer (DHT) that dynamically configures layers and heads per sample using contextual bandit methods, achieving up to 74% computational savings with minimal accuracy loss.
Transformer requires a fixed number of layers and heads which makes them inflexible to the complexity of individual samples and expensive in training and inference. To address this, we propose a sample-based Dynamic Hierarchical Transformer (DHT) model whose layers and heads can be dynamically configured with single data samples via solving contextual bandit problems. To determine the number of layers and heads, we use the Uniform Confidence Bound while we deploy combinatorial Thompson Sampling in order to select specific head combinations given their number. Different from previous work that focuses on compressing trained networks for inference only, DHT is not only advantageous for adaptively optimizing the underlying network architecture during training but also has a flexible network for efficient inference. To the best of our knowledge, this is the first comprehensive data-driven dynamic transformer without any additional auxiliary neural networks that implement the dynamic system. According to the experiment results, we achieve up to 74% computational savings for both training and inference with a minimal loss of accuracy.