I recently trained a Text Classification (Single-label) Model using Google Cloud AutoML, and it took 4 hr 48 min to run. How can I speed up this process? Are there certain parameters that I can change so that the training runs faster? There were only 2,825 total items, with only a couple sentences per item. I've trained the same data with Microsoft, and it only took ~45 minutes. Why is it taking so long for Google? Thanks!
Referencing the Pricing for AutoML models page:
Vertex AI uses predefined machine configurations for Vertex AutoML models whereas for Custom-trained models: You can choose a custom configuration of selected machine types
Is there any way to configure the compute resources for Vertex AutoML via an MLOp Workflow or Vertex Pipeline? Or can this only be adjusted via a custom training method?
