Should We Still Pretrain Encoders with Masked Language Modeling?
Paper
•
2507.00994
•
Published
•
80
MLM versus CLM for NLP tasks
Related paper: "Should We Still Pretrain Encoders with Masked Language Modeling?"