search

e-apostolidis / PGL-SUM

A PyTorch Implementation of PGL-SUM from "Combining Global and Local Attention with Positional Encoding for Video Summarization" (IEEE ISM 2021)
Other
84 stars 32 forks source link

Could you give me example in training #4

Closed tingchihc closed 2 years ago

tingchihc commented 2 years ago

I know this line is to train model. python model/main.py --split_index N --n_epochs E --batch_size B --video_type 'dataset_name' However, I do not how to setup these parameters. Can you give me the sample ?

thanks,

mpalaourg commented 2 years ago

Hello @ting-chih, the reported results are by running the run_{dataset}_splits.sh script (e.g., for SumMe). We only care about those parameters, because we have set the default values of the argparse in the configs.py script with the appropriate ones. However, you can use any of these parameters for your experiments (more info in the README file) as an argument (with -- in front).

The structure of the code assumes that you have a folder for the project (for example named PGL-SUM) and each script will be executed from this folder (w/o cd-ing in and out of the given folders). Hope this helps!

mpalaourg commented 2 years ago

Closing due to inactivity. Feel free to re-open the issue or create a new one, for furthermore questions and queries.