Open jeongwonmin opened 7 years ago
Ah yes. I need to update the readme.
You must prepare the dataset with the flag "--tframes 100”.
From Corey: After the dataset has been prepared for the CDNN run train_mlp_conv_script.py (not train_mpl_script.py) with the file mlp_rlu_conv2.yaml configuration file and your newly prepared train/valid/test configuration fold.
Thank you for your comment! It started working after I had prepared the dataset with the flag "--tframes 100".
I read your paper "Deep learning and music adversaries" and I am doing some experiments with your source code. I tried to train my dataset with "mlp_rlu_conv2.yaml" after preparation by using "prepare_dataset.py", but it returns an error message "The value of a Conv2DSpace batch must be 4D".
I also looked for the file "audio_dataset2d", but there was no "audio_dataset2d.py".
What do I have to do before training with the convolutional neural network?