Greg

Monorepo project will provide - Shared code and visibility - Atomic changes

Merge and rework [jcloud](https://github.com/jina-ai/jcloud)

Need to remove dependency on ` jina-hubble-sdk ` from `docarray` as this ties directly into Jina Cloud Service. Primary offending class: [pushpull.py]( https://github.com/docarray/docarray/blob/52a07c0da4c99d03c87b383dbc63ce45ea8a1627/docarray/array/mixins/io/pushpull.py)

Core framework should support JAX framework out off the start.

In `find_model_using_name` there is an issue with the dynamic imports. ``` model_filename = "models." + model_name + "_model" modellib = importlib.import_module(model_filename) ``` Workaround is to have them map directly ```...

Add Inte https://medium.com/pytorch/accelerate-pytorch-with-ipex-and-onednn-using-intel-bf16-technology-dca5b8e6b58f

Implement metrics and APDEX (Application Performance Index)

Core framework should support TensorFlow out off the start.

`OverlayProcessor` always get assigned default GPU ID 0. This needs to be configurable as when running on Multi-GPU this should use the correctly assigned one. Source: ``` marie.overlay.overlay.OverlayProcessor @staticmethod def...

Based on the discussion here [How to cluster words into semantic entities, when performing information extraction? #92](https://github.com/microsoft/unilm/issues/923) need to add support for vertical named entity aggregation.