You are using a model of type t5 to instantiate a model of type GraphLlama. This is not supported for all configurations of models and can yield errors.
When I test the Pretrained-model ：Fastchat-t5 model to demonstrate the GraphGPT model, some issue are faced with this problem. Can the author give some suggestions to address this one?
You are using a model of type t5 to instantiate a model of type GraphLlama. This is not supported for all configurations of models and can yield errors.
When I test the Pretrained-model ：Fastchat-t5 model to demonstrate the GraphGPT model, some issue are faced with this problem. Can the author give some suggestions to address this one?