No such file or directory: '../../data/splits/split-0/all_train_dict.pickle' when reproducing experiments

[expectopatronum]

Hi, I tried to reproduce your experiments as described in the README of your repository: python scripts/baseline/main.py --mode 'TRAIN'

I get the following error: No such file or directory: '../../data/splits/split-0/all_train_dict.pickle' Did I miss running some preprocessing script? Thanks for your help! Best regards, Verena

[minzwon]

Hi,

From scripts/baseline/get_npy.py, you can find the run function. That will generate the pickle files that you need.

[expectopatronum]

Hi, thanks for your quick response. The script contains a hardcoded path '/home/minz/Developer/data/jamendo/npy'. What should I replace this with? I have downloaded the following data: python scripts/download/download.py --dataset autotagging_moodtheme --type audio /home/verena/shared/MTG-Jamendo/audio/ --unpack --remove which does not contain a npy directory. I was wondering if I might need to download the preprocessed melspecs for this script to work but I wanted to verify it first before downloading that much data. Thanks and best regards, Verena

[minzwon]

Sorry for the late response. I missed the notification. You should put your path to the spectrogram. You can download the spectrograms or you can do your own preprocessing using the audio files.

I updated the get_npy.py so that you can simply run: python get_npy.py run 'your_path_to_spectrogram_npy'