PKU-YuanGroup / LanguageBind

【ICLR 2024🔥】 Extending Video-Language Pretraining to N-modality by Language-based Semantic Alignment
https://arxiv.org/abs/2310.01852
MIT License
723 stars 52 forks source link

Are some of these models interchangeable? #55

Open felmoreno1726 opened 5 months ago

felmoreno1726 commented 5 months ago

For example, I wonder if I train an LLM model using one LanguageBind/LanguageBind_Video_FT, LanguageBind/LanguageBind_Video, or LanguageBind/LanguageBind_Video_V1.5_FT. Can I later swap the Video Encoder for one of the other ones? Or would I need to retrain said LLM with a different encoder if I wish to swap the encoder? Should these give approximately similar results?