This message was deleted.

Hello Jawher, did you look at this part of doc: https://docs.bentoml.org/en/latest/concepts/runner.html#resource-allocation It may help if you did not see.

So if I understood correctly. I just need to create a configuration.yml file and specify 'cpu: 1000m' if I want to limit a specific runner to 1GB of RAM?

At first I thought same think, but I think that's not the case. I am not sure but I think when you set 1000m means 1CPU. I think docs are not clear on it but when I look at the source code I found: https://github.com/bentoml/BentoML/blob/dcdf3765a6eb6ca8b871280d5719cc580a956d6e/src/bentoml/_internal/resource.py#L77 .

Yep, you're right

I am not sure if this only for deployment with yatai but I found a deployment config you may investigate more. https://github.com/bentoml/BentoML/blob/632141f197e7e3415781f7545039f318975238de/examples/inference_graph/deployment.yaml#L25

it seems to be working

Hi there, right now we haven’t implemented a mem resource yet.

@Jawher Ben Abdallah can you tell us more about the three models deployed? What frameworks are they? Are you implementing custom runners?

Hey @Sean, the three models are two simple CNNs and YOLOv7. They were all exported to ONNX and I use the BentoML default runner to call them.