🔫 Zero-Shot and Few-Shot Classification With SetFit




  • The code below will load the banking customer requests dataset from the Hub, encode the text field, and create the vectors field which will contain only one key (mini-lm-sentence-transformers). For the purposes of labelling the dataset from scratch, it will also remove the label field, which contains the original intent labels. [ ]: Copy to clipboard (View Highlight)
  • SetFit’s approach to zero-shot is to create a synthetic dataset of training examples, which is different from other approaches (e.g., transformers zero-shot pipelines) where “templated” examples with label names are used at inference time. (View Highlight)