This message was deleted.
# ask-for-helps
Slackbot
07/18/2023, 6:36 PMThis message was deleted.
👍 1
v
Visaal Ambalam
08/04/2023, 6:26 PMI’m going to try and copy and modify what’s here
https://github.com/bentoml/BentoML/tree/2c627c3b3aa4beaf374c68a80c111d1bc2f97fdc/examples/pytorch_mnist/tests
Visaal Ambalam
08/04/2023, 6:27 PMIt’s odd that it seems to launch a test bento server when trying to run the unit tests.
Shouldn’t there be a way to supply mock/test dependencies to the inference service to be able to write unit tests?
c
Chaoyu
08/06/2023, 9:26 PMwe will add some more example for unit testing, you’re right that example was more of an e2e test of a Bento
gratitude thank you 1
Chaoyu
08/06/2023, 9:26 PMJust curious, in your Unit tests, do you expect to load the actual model? or ideally you will mock all the runner calls?
v
Visaal Ambalam
08/07/2023, 3:59 PMYeah, ideally we’d have few e2e unit tests with the actual model and others with a mock runner/model.
Visaal Ambalam
08/07/2023, 4:16 PMHas anyone had success defining an inference service without putting the
svcmodelrunnerc
Chaoyu
08/07/2023, 4:23 PMyes you should be able to do that
Chaoyu
08/07/2023, 4:23 PMI’m planning to add some more unit test for this example project for references btw https://github.com/bentoml/CLIP-API-service
Chaoyu
08/07/2023, 4:24 PMThis product have model and runners defined I. Different files too