Is there any way to use T-NER without downloading the model every time?

[MrDevCop]

I understand downloading the model once. It gets stored in the cache and is available for the user in that session. But I want to know how to use the method TransformerNER() with a specified model without downloading it every time, even if a session renews. For that to happen I believe I will have to download the model files in the system and somehow point the method TransformerNER to the required paths.
Any help would be much appreciated.

Background: We are trying to use tner library to identify entities in a client's dataset. However, we can access the dataset via a VM on which internet access is not available. But we can transfer files and folders to the VM. So we were thinking of a way to download the models on our own systems and then transfer the required files for TransformerNER method manually to the VM.

PS: We only need to use TransformerNER method.

[asahi417]

Thanks for using TNER!

We store models on huggingface model hub https://huggingface.co/, and when you load a model on TNER, it should be cached under ~/.cache/huggingface/, and it will use the cache from next time, unless it has been updated. So, I assume you can mount that huggingface cache folder to the VM when you launch it, and TNER should work offline.

[asahi417]

I use a computing cluster for TNER model training, which also has no internet connection, so the above steps are what I actually do in my environment.