BentoML

Screen Shot 2022-10-24 at 11.33.22 AM.png

hello <@U044HEP2FLZ> Is there a particular reason, you want to use the file path instead of the tag?

typically, you would save the model with bentoml and then when you want to load, you can just have `model = bentoml.keras.load_model('my_model:version')`.

yeah that always points to the models folder in bentoml, i would like to point it to a folder inside the projevct

Screen Shot 2022-10-24 at 11.39.28 AM.png

this always points to the models folder in users/username/bentoml/models/modelname

i want to save it in a different location

we are trying to save all the models of 1 customer in 1 folder, for example customer1: model 1, model 2 etc customer2: model 1, model 2 etc

so we want to be able to load from different locations

hi there, when you specify the version inside `load_model`  it ensures that you will load the model from that version only.

Ideally, I suggest you not to load model this way, as the model saved inside bento is exactly the same with the one that you use `load_model`

For managing multiple models, I suggest using `labels` as demonstrated here <https://docs.bentoml.org/en/latest/concepts/model.html#save-a-trained-model> to differentiate among different models

hi <@U01BMCJ29LN> i am trying to pass the model ref to service and i am seeing the following error

You should always create a runner and then pass the runner to the service

Model ref are recommended to use when creating a custom runner