Awni Hannun
Awni Hannun
@NripeshN are you planning to come back to this PR?
@mlaves this looks really nice!! I think the main question is what the purpose of the `ConvolutionTranspose` primitive is? I didn't look to carefully, but it looks like we should...
@mlaves you are planning to come back to this? Would be great to land it.
I remove the tranpose flag and attempted to resolve the incorrect gradient w.r.t. the `weight` when the `flip` argument is `True`. I don't think all the gradients are correct yet...
We don't have it, but you can certainly do a binary search with existing ops. Here's something for 1D arrays that works, it should be fairly straight-forward to support an...
I'm open to adding a little binary search implementation like that into MLX to support `searchsorted`. We could use the above as a starting point.
Another option is to do something like the following. It's linear in `a` but probably quite a bit faster, especially for small arrays, since it's far few operations: ```python def...
By all means
I would compare ```python def searchsorted(a, b): axis = 0 size = a.shape[axis] steps = math.ceil(math.log2(size)) upper = size lower = 0 indices = mx.full(b.shape, vals=size//2, dtype=mx.uint32) for _ in...
There is no update. You can use the above for now. If nyone wants to work on the issue it's open... we'll try to get to it otherwise when we...