This message was deleted.

Hiya, yes, you can do this by setting

BENTOML_HOME

.

Thanks, @sauyon. I will try that. Just an FYI : With Truss, I can specify the target directory while saving the model.

tr = mk_truss(model, target_directory="sklearn_truss")

If you need the model in another place or machine, I'd suggest exporting the model with

bentoml export

. Setting the home is in case you'd prefer your local machine's data to be in a different place.

Where can I see all the list of environment variables, we can set for bentoml? Do we have any cli command to list them?

Unfortunately I don't think we have good documentation for this. Was there anything in particular you wanted to do?

Just wanted to know. Thanks again for your help.

maybe we can take inspiration from

go env

we can have a

bentoml envars

?

I think we should try to print the relevant env vars in

bentoml env

, but this also just needs better documentation, I think.

right. I guess the current

bentoml env

can become

bentoml info

?

I don't think that's necessary? We can just have it show the relevant env vars first or in a different section, would make it easier for us to debug anyway.