Pretrained Model download archives

Looking through the documentation, I don’t see anything about how long pre-trained models are expected to be available for download.

Is the plan to make a released pre-trained model to be available indefinitely?

For example, if I build our provisioning process to install the following model, should this URL always exist:

https://artifacts.opensearch.org/models/ml-models/amazon/neural-sparse/opensearch-neural-sparse-encoding-doc-v1/1.0.1/torch_script/neural-sparse_opensearch-neural-sparse-encoding-doc-v1-1.0.1-torch_script.zip

Or should I archive the models we plan on using locally and always install from our locally hosted models to ensure that the models are always available?

I just want to make sure that if I point to the public URLs, those should always be expected to return the model files even when they’ve been replaced with newer versions.

Yeah, the URL will always exist. We might release a newer version later, but that shouldn’t affect the current version.

@dhrubo Thanks. I just wanted to confirm that if my provisioning scripts are based on the URLs, they won’t break in the future.

Yeah, they won’t break in the future!