KeyError: 'video_fnames'

[Forainest789]

I followed the readme to run the training scripts And I could run the

bash scripts/run.sh --train 

it could train and predict but in the last epoch after saving and loading the best kept it do a more evaluation and will have the following error

Moment Retrieval loss: 0.0235
Step Captioning loss: 0.6253
Saved ./checkpoints/hirest_joint_model/moment_retrieval_epoch_049.json
Saved ./checkpoints/hirest_joint_model/moment_segmentation_epoch_049.json
Saved ./checkpoints/hirest_joint_model/step_captioning_epoch_049.json
Model saved at ./checkpoints/hirest_joint_model/LAST.pth
Best Epoch:  5
Model loaded from  ./checkpoints/hirest_joint_model/BEST
Moment_retrieval Prediction:   0%|                                              Moment_retrieval Prediction:   1%|▎                                             Moment_retrieval Prediction:   4%|██                                            Moment_retrieval Prediction:   6%|███▍                                          Moment_retrieval Prediction:  10%|█████                                         Moment_retrieval Prediction:  12%|██████▍                                       Moment_retrieval Prediction:  15%|███████▊                                      Moment_retrieval Prediction:  17%|█████████▏                                    Moment_retrieval Prediction:  20%|██████████▌                                   Moment_retrieval Prediction:  22%|███████████▉                                  Moment_retrieval Prediction:  25%|█████████████▎                                Moment_retrieval Prediction:  28%|██████████████▉                               Moment_retrieval Prediction:  31%|████████████████▋                             Moment_retrieval Prediction:  35%|██████████████████▎                           Moment_retrieval Prediction:  38%|████████████████████                          Moment_retrieval Prediction:  41%|█████████████████████▋                        Moment_retrieval Prediction:  44%|███████████████████████▍                      Moment_retrieval Prediction:  47%|█████████████████████████▏                    Moment_retrieval Prediction:  50%|██████████████████████████▌                   Moment_retrieval Prediction:  53%|███████████████████████████▊                  Moment_retrieval Prediction:  55%|█████████████████████████████▏                Moment_retrieval Prediction:  58%|██████████████████████████████▌               Moment_retrieval Prediction:  60%|███████████████████████████████▉              Moment_retrieval Prediction:  63%|█████████████████████████████████▎            Moment_retrieval Prediction:  66%|██████████████████████████████████▎           Moment_retrieval Prediction:  69%|████████████████████████████████████          Moment_retrieval Prediction:  72%|█████████████████████████████████████▋        Moment_retrieval Prediction:  76%|███████████████████████████████████████▎      Moment_retrieval Prediction:  79%|█████████████████████████████████████████     Moment_retrieval Prediction:  82%|██████████████████████████████████████████▋   Moment_retrieval Prediction:  85%|████████████████████████████████████████████▎ Moment_retrieval Prediction:  88%|██████████████████████████████████████████████Moment_retrieval Prediction:  92%|██████████████████████████████████████████████Moment_retrieval Prediction:  95%|██████████████████████████████████████████████Moment_retrieval Prediction:  98%|██████████████████████████████████████████████Moment_retrieval Prediction: 100%|████████████████████████████████████████████████████| 156/156 [00:04<00:00, 35.54it/s]
moment_segmentation <torch.utils.data.dataloader.DataLoader object at 0x7f9a56397a30>
Moment_segmentation Prediction: 0it [00:00, ?it/s]
Traceback (most recent call last):
  File "/hy-tmp/HiREST/run.py", line 903, in <module>
    main_worker(0, args)
  File "/hy-tmp/HiREST/run.py", line 854, in main_worker
    trainer.train()
  File "/hy-tmp/HiREST/run.py", line 497, in train
    test_moment_segmentation_results = self.evaluate(self.test_moment_segmentation_loader, has_target=False)
  File "/hy-tmp/HiREST/run.py", line 840, in evaluate
    results = self.predict(loader, has_target=has_target)
  File "/hy-tmp/HiREST/run.py", line 687, in predict
    assert len(results['tasks']) == len(results['video_fnames']) == len(results['prompts']) \
KeyError: 'video_fnames'

it seems like there is a closed issue similar to mine. I have tried that method, but it didn't work