Dependencies

[justinormont]

Modifications:

• Reordering torch first in dependency list, as it's required by subsequent items
• Add missing fairseq dependency

Basic build/install steps:

git clone https://github.com/lucidrains/musiclm-pytorch
cd musiclm-pytorch
conda create -n musiclm-pytorch python=3.10
conda activate  musiclm-pytorch
python setup.py install

Granted, I think the project is incomplete still, so won't work yet. The above is just to get it to build/install.

[gijzelaerr]

as far as i know, the order in install_requires doesn't matter. The list is fully scanned and then the dependency solver builds a solution, installing all requirements and their dependencies. Also, I can install and import the lib fine without fairseq listed in the requirements.

[justinormont]

I was previously under the same assumption. Without the reordering, the install didn't work for me, giving errors of missing PyTorch.

Are you working under Conda? I assume you don't have a system level pip pages installed (e.g. PyTorch)? I expect it won't matter until runtime, though I'm on macOS.

[StudiousDog]

for the trainers used in this project there is an input argument "folder" which takes a path to the folder of sound dataset files, which I have no idea what it is, looks like I need a sound dataset for the trainer to work. Does anyone know what that is/where to find it?
musiclm-pytorch\musiclm_venv\lib\site-packages\sklearn\base.py:318: UserWarning: Trying to unpickle estimator MiniBatchKMeans from version 0.24.0 when using version 1.2.2. This might lead to breaking code or invalid results. Use at your own risk. For more info please refer to:
https://scikit-learn.org/stable/model_persistence.html#security-maintainability-limitations
warnings.warn(
Traceback (most recent call last):
File "C:\ai\musiclm-pytorch\scripts\main.py", line 9, in
audio_lm = train_transformers(quantizer)
File "C:\ai\musiclm-pytorch\scripts\train_transformers.py", line 37, in train_transformers
semantic_trainer = SemanticTransformerTrainer(
File "<@beartype(audiolm_pytorch.trainer.SemanticTransformerTrainer.init) at 0x21350b512d0>", line 116, in init
File "C:\ai\musiclm-pytorch\musiclm_venv\lib\site-packages\audiolm_pytorch\trainer.py", line 592, in init
self.ds = SoundDataset(
File "<@beartype(audiolm_pytorch.data.SoundDataset.init) at 0x21350b1bc70>", line 116, in init
File "C:\ai\musiclm-pytorch\musiclm_venv\lib\site-packages\audiolm_pytorch\data.py", line 49, in init
assert len(files) > 0, 'no sound files found'
AssertionError: no sound files found