panns_transfer_to_gtzan

panns_transfer_to_gtzan copied to clipboard

数据集链接失效

2

您好，感谢您的出色工作。当我想复现您的工程时，发现数据集下载链接http://marsyas.info/downloads/datasets.html已失效。您可以再重新发下吗？

wusuowei11

about runme.sh

2

hey, I want to ask you about runme.sh. `python3.6 pytorch/main.py train --dataset_dir=$DATASET_DIR --workspace=$WORKSPACE \ --holdout_fold=1 --model_type=$MODEL_TYPE --pretrained_checkpoint_path=$PRETRAINED_CHECKPOINT_PATH \ --freeze_base --loss_type=clip_nll --augmentation='mixup' --learning_rate=1e-4 --batch_size=16 \ --few_shots=10 --random_seed=1000 --resume_iteration=0 --stop_iteration=10000 --cuda` Can...

dopc

I am confused about the calculation clip_nll loss. If my predict : clipwise_output = [-1.2 , -2.3, -0.5] target = [0, 0, 1] torch.mean(clipwise_output * target) = torch.mean([0, 0, -0.5])...

YuGuiwe

I am unsure what I can permissively use this code for without license file. I would appreciate if this is added for clarity. Thanks.

nandan-sridhar

problem with audio tagging when use inference code

2

Hello Thanks for sharing codes I appreciate your consideration In this readme file, you haven't written anything about how we can inference a new file and tag it with the...

aliakbartaghizadeh

HI, i'm trying to do transfer learning for Sound event Detection but it is not working as expected, it is training for the same values  is there any other...

rohithkodali

Thank you for your great work! when I train my audio dataset，I change the sample_rate=16k，and other params keep unchanged. I can sucess to train ,and the final validate accuracy is...

Keyird