Is Grappler's autoparallel optimizer un maintained?

Is this optimizer currently unmaintained?

There is little docs about this. Also, it only applies to a few gradient ops from the code in tensorflow/core/grappler/optimizers. The autoparallel pass is quite straightforward and could only apply to few scenario, I think?

Could someone offer a little more information about this optimizer?

1 Like

I think that the developmemt of these optimizations is going to migrate over MLiR

1 Like