PISCES is a pre-trained many-to-many summarization model that learns language modeling, cross-lingual ability and summarization ability through a designed three-stage pre-training.

This model is introduced by Towards Unifying Multi-Lingual and Cross-Lingual Summarization (To appear in ACL 2023 main conference)

tokenizer = MBart50Tokenizer.from_pretrained('Krystalan/PISCES')
model = MBartForConditionalGeneration.from_pretrained('Krystalan/PISCES')