Closed FeatherZhong closed 3 years ago
Hi, I've added a new code in stage2: convert_file_format.py
Please run the file to convert the file format. ("vist_scored_terms_6_path.json" to "vist_scored_terms_6_path_for_stage3.json". )
Thus, the term_path will be modified to vist_scored_terms_6_path_for_stage3.json. e.g.) python inference_added_story.py -model save_model_finetuneLDPE_pretrain_vist/trained_accu_55.397.chkpt -device 3 -positional LDPE -term_path ../../stage2/vist_scored_terms_6_path_for_stage3.json
also, I've updated the code to download the saved checkpoint for the model in stage2/data/download_big_data.sh.
Or you could get it by typing: wget https://doraemon.iis.sinica.edu.tw/KE_VIST_AAAI2020/trained_accu_55.397.chkpt
Thank you very much for your help.This code worked!
No problem! Glad to help!!!
Thank you very much for your help.This code worked!
Hi,FeatherZhong Now,I'm replicating this code,but i encounter many difficuties.The biggest difficulty is that the download links invalid.Saw you haved run this code successful.I wonder if it is convenient to provide me with a complete code.Thanks a lot!
Hi, I am tring to generation story in stage 3 using the command you provide below:
python inference_added_story.py -model save_model_finetuneLDPE_pretrain_vist/trained_accu_56.156.chkpt -term_path ../../stage2/vist_scored_terms_6_path.json -device 1 -positional LDPE -insert 0 -relation 0
but I get some errors and found that they came from the wrong format of the file vist_scored_terms_6_path.json generated from stage2. In the ROCAddTermsetDataset which using by the data loader the dialogs loaded from vist_scored_terms_6_path.json is a dictionary, not a list, so it cannot be indexed at here, and I can't see the predicted_term_seq field in the file mentioned here . I would appreciate it if you could help me to solve this problem.