Base.pkl file is missing while trying to use MSP60k

[asad1416]

I am trying to train model with MSP60k but it gives that .pkl file missing. I figure out that I have to generate file using .py files. But Now I am stuck it is asking base.pkl file which is not present in whole directory. Please guide a step step approach to make this dataset ready for training.

split_1 Traceback (most recent call last): File "/UNITY-NFS/Data_B/OpenPAR/SUBMIT/msp_withMS.py", line 141, in generate_pkl_MS(save_dir, exist_pkl_path=exist_pkl_path, only_change_root=False) File "/UNITY-NFS/Data_B/OpenPAR/SUBMIT/msp_withMS.py", line 99, in generate_pkl_MS exist_pkl = pickle.load(open(exist_pkl_path, 'rb+')) FileNotFoundError: [Errno 2] No such file or directory: '/UNITY-NFS/Data_B/OpenPAR/SUBMIT/split/base.pkl'

[asad1416]

Currently the dataset have 3 .pkl files dataset_random.pkl, dataset_ms_split1.pkl and dataset_ms_split2.pkl. I tried all three files giving in local.py. But I am getting follwing error File "/UNITY-NFS/Data_B/OpenPAR/MSP60K_Benchmark_Dataset/LLM-PAR/train.py", line 261, in main(args) File "/UNITY-NFS/Data_B/OpenPAR/MSP60K_Benchmark_Dataset/LLM-PAR/train.py", line 30, in main train_set = MultiModalAttrDataset(args=args, split=args.train_split, transform=train_tsfm) #得到训练数据集 File "/UNITY-NFS/Data_B/OpenPAR/MSP60K_Benchmark_Dataset/LLM-PAR/dataset/AttrDataset.py", line 36, in init self.sentences = dataset_info.sentences AttributeError: 'EasyDict' object has no attribute 'sentences'

[1125178969]

if you want run the LLMPAR, you need generate the Template.pkl use the scripts for the corresponding dataset in MSP60K_Benchmark_Dataset/LLM-PAR/PAR_Template_Scripts

[1125178969]

https://github.com/Event-AHU/OpenPAR/issues/28#issue-2512900555 you can set the exist_pkl_pathas None