Zhuang Li
Hi Have you solved this problem? I am facing the same issue that I use BERT as an encoder, the performance drops significantly.
yes met the same issue. While I see some works report that they use BERT as encoder and the RNN as decoder and got good results. Don't know why.
same issue. It shows nothing. Have you solved your problem? thanks.
Yeah. I solved the issue. It is because of the data format. I am using the code to train llama for StarCoder. Then I fixed the error by properly preprocessing...
@wangqian621 We have a new work published on ACL2023 for this. The model now reaches the SOTA. If you are interested, please refer to https://github.com/zhuang-li/FactualSceneGraph
@zyong812 Hi, our new scene graph parser can well handle this situation. The paper has been published in ACL 2023. Welcome to have a try. https://github.com/zhuang-li/FACTUAL
The same question, thx. I am trying to replicate your experiments as my baseline. But I don't even know how to process the data. The original dataset is in json...
never mind. I figured out myself