cannot download and execute a model locally

[zfarrell13]

I have the appropriate models for htdemucs_ft downloaded and in a folder:

/models
     /92cfc3b6-ef3bcb9c.th
     /04573f0d-f3cf25b2.th
     /d12395a8-e57c48e6.th
     /f7e0c4bc-ba3fe64a.th

I am trying to execute them locally with the command:

python3 -m demucs.separate -o separated_sources --repo .../models -n htdemucs_ft --mp3 --mp3-bitrate=320

but i keep getting the error:

FATAL: htdemucs_ft is neither a single pre-trained model or a bag of models.
Command failed, something went wrong.

I see repo.py is responsible for handling --repo arguments via the command line. is there any specific structure that the models must be saved in?

could anyone assist please?

[zfarrell13]

for anyone else that will be trying this, you can figure out which models you need using the collab on the readme, defining the model name in the "model =" parameter.

The demucs repository lists the models needed for each model name in the remote folder. buttt the full url needed to download a model or a bag of models is not noted. Use the collaab and see what downloads to get the full URLs.

download all models locally from the links, you also need the yaml file associated to the model name from the remote folder in the directory you place all of the downloaded models into. after that a --repo argument followed by the path to the local model directory can be run via the command line with the other execute arguments in order to run the process locally. for anyone that had a hard time or is not seasoned in reading the raw code to figure this out, feel free to msg me.

[zfarrell13]

closing