Explains why large transformer inference is costly in time and memory
Covers the two main bottlenecks behind inference challenges
Includes a section on distillation for model compression
Key quotes
“The extremely high inference cost, in both time and memory, is a big bottleneck for adopting a powerful transformer for solving real-world tasks at scale.”
Transcript
[Updated on 2023-01-24: add a small section on Distillation .] Large transformer models are mainstream nowadays, creating SoTA results for a variety of tasks. They are powerful but very expensive to train and use. The extremely high inference cost, in both time and memory, is a big bottleneck for adopting a powerful transformer for solving real-world tasks at scale. Why is it hard to run inference for large transformer models? Besides the increasing size of SoTA models, there are two main factors contributing to the inference challenge ( Pope et al. 2022 ):
