Ahmadzei's picture
added 3 more tables for large emb model
5fa1a76
Speech and vision models have their own pretraining objectives.