Batuhan Taskaya

@alamb @Dandandan @mingmwang I've created the meta/epic issue as we discussed

I'd be happy to start one, and if anyone is interested I can also give write access (shoot me your google emails at `[email protected]`).

It should be publicly accessible now: https://docs.google.com/document/d/1M4mmV7KA1LSj-D-WJA338B4ydlm-8A8D5OPuDE5_SD4/ (also pinning this to the issue) It is an overall discovery of the stuff we are doing right now and how they can...

Thanks @alamb! I'll also try to talk a bit more about it with real-world examples in [tomorrow's meeting](https://arrow.apache.org/datafusion/contributor-guide/communication.html#sync-up-video-calls) from scratch (if we would have the time for that in this...

> Is there a scenario which you want to fallback to torch eager( because it will be slow). Torch's `dynamo.run` actually explains it pretty well > In some cases, you...

@strint this still makes sense for us where we are compiling pipelines that might not be officially supported by onediff yet, e.g. for AnimateDiffVideo2Video pipeline, I was able to get...

Essentially I'd like to keep the graph itself in GPU since it is a single graph, but offload the weights back and forth between GPU and RAM and re-use the...

Please see https://github.com/python/core-workflow/issues/363 for the previous discussion in this matter.

CC: @ewdurbin

hey @jeremyhylton, i am currently a bit busy with non-oss stuff so if you'd like feel free to take over the PR (and close this one). whatever is the most...