Labelbox ：2022年自动化标注指南报告（英文版）

Developing and maintaining a performant service using supervised machine learning requires a huge amount of data. Training large models on large datasets creates several challenges, including: 1. Working with distributed graphics processing units (GPUs) and specialized hardware like tensor processing units (TPUs) 2. Figuring out how to run potentially time- and cost-intensive experiments to validate that your changes do improve model performance 3. Labeling the vast amount of data necessary to t