_target_: text_recognizer.model.mammut.LitMaMMUT
max_output_len: 682
caption_loss_weight: 1.0
contrastive_loss_weight: 1.0