Feature request - Offline use of model

[Phil1108]

At the moment it is nearly impossible to create a docker container that works offline (without internet access).
Even if you include this line during docker build:

RUN python -c "import torch; torch.backends.quantized.engine='qnnpack'; torch.hub.load(repo_or_dir='snakers4/silero-models', model='silero_te', force_reload=True)"

During execution of the docker container (without internet) you load it locally:

torch.hub.load(repo_or_dir='/root/.cache/torch/hub/snakers4_silero-models_master', model='silero_te', source='local', force_reload=False)

Then you have the problem that the hubconf.py is called again (and fails due to no internet access) and it tries to download the files in hubconf.py Lines 21, 49, 101, even though they already exist.

So my suggestion would be to also includes checks in the Lines 21,49,101 to check if the file already exists locally and if yes then skip it (like done in Line 114)

Any reasons against that?

[snakers4]

Hi,

Then you have the problem that the hubconf.py is called again (and fails due to no internet access) and it tries to download the files in hubconf.py Lines 21, 49, 101, even though they already exist.

Many thanks for catching this, this is a valid edge case. This is not intentional.

Originally I believed that people using these models locally would just override the hubconf.py with their own init functions. At that moment I needed a list of models to be loaded before doing the rest, so I just used the simplest dumbest solution and never gave it a second thought.

So my suggestion would be to also includes checks in the Lines 21,49,101 to check if the file already exists locally and if yes then skip it (like done in Line 114)
Any reasons against that?

I do not know why I overlooked this method hub_dir = torch.hub.get_dir().
Since you are already using pytorch.hub cache, maybe just load the .yml file from hub_dir, if it exists?

A small PR would be greatly appreciated.

[axenov]

@snakers4 I have added a small fix to the issue.

[snakers4]

Just a small question, why use __file__ instead of hub_dir = torch.hub.get_dir()?

[axenov]

torch.hub.get_dir() returns path to the whole torch.hub cache and not only cache of silero-models. You used __file__ to get the right directory bellow: https://github.com/snakers4/silero-models/blob/master/hubconf.py#L69

[snakers4]

Merged #111
Checked that all of the main examples work (STT, TTS, TE)

[snakers4]

Please verify, that it runs locally with your build now and close

[Phil1108]

Yeah works like a charm. Thanks @snakers4 and @axenov