WaLDORf: Wasteless Language-model Distillation On Reading-comprehension
This addresses the resource-intensive deployment of large language models for natural language understanding tasks, offering a practical solution for production environments.
The paper tackles the problem of deploying large language models efficiently by proposing WaLDORf, a task-specific hybrid model that achieves state-of-the-art inference speed and higher accuracy than previous distilled models.
Transformer based Very Large Language Models (VLLMs) like BERT, XLNet and RoBERTa, have recently shown tremendous performance on a large variety of Natural Language Understanding (NLU) tasks. However, due to their size, these VLLMs are extremely resource intensive and cumbersome to deploy at production time. Several recent publications have looked into various ways to distil knowledge from a transformer based VLLM (most commonly BERT-Base) into a smaller model which can run much faster at inference time. Here, we propose a novel set of techniques which together produce a task-specific hybrid convolutional and transformer model, WaLDORf, that achieves state-of-the-art inference speed while still being more accurate than previous distilled models.