tonyhu

Would like to have this available in PyPI as well for easy installation

This is an issue due to ray setup logics for windows machine at [here](https://github.com/ray-project/ray/blob/releases/2.9.3/python/setup.py#L234). We need to wait for ray to release this [fix](https://github.com/ray-project/ray/pull/42097) for final resolution. A temp mitigation...

> Could you please also test try bigger models like Flan-t5-XXL or XL? Currently it seems weird with deepspeed on, looks like a perfect candidate for the deepspeed trainer. Thanks....

One interesting observation is that the docs are built fine with torch 2.2 at [here](https://auto.gluon.ai/dev/tutorials/multimodal/object_detection/quick_start/quick_start_coco.html)