5fa1a76
1
2
3
When using pre-trained models for inference within a [pipeline], the models call the PreTrainedModel.generate() method that applies a default generation configuration under the hood.
