label_text_file = '/data/vision/torralba/crossmodal/soundnet/lmdbs/train_frames4_%04d.txt'

[avis-ma]

Hello，thank you very much for sharing the code and the dataset. I am a new one in audio_visual. I want to reappear the source code. when I read the code ,I find I do not have the file "label_text_file" in main_train.lua. If i want to train the model by myself, i have to have the raw mp3s(359GB), and the image features(88GB), but i dont know how to get the label_text_file. if everyone knows, or everyone had re-trained the model by yourself, please give me some advise or some experience. thank you very much