Switch to Multimodal Reader
Closed this issue · 0 comments
lwaekfjlk commented
- [] Change the current t5-base model into multimodal based t5 like VL-T5 model.
- [] Test whether VL-T5 model perform better or worse than t5-base under the pure text setting
- [] Handling the multi-hop problem
- [] Test whether BLIP model perform better or worse than t5-base under the pure text setting