Reason for end of TensorFlow back-end support

[grebdioZ]

I only recently realized that support for the TensorFlow back-end was dropped with version 25.03. I have totally missed the announcement, but I admit I did not follow release notes or newsletters closely.

Anyway, no reason was given in the release notes, but I am sure the reason for such a significant decision must be given somewhere? Or is there at least a discussion someone can point me to?

Also, are we likely to see ONNX or PyTorch support be dropped soon, too?

[grebdioZ]

Bumping.

While I would love to know the reason for the discontinuation, I can probably do nothing about it anyway.

But the question on what your plans are with ONNX and PyTorch support is somewhat urgent, as we need to migrate to one of those, if we want to keep using Triton (probably pytorch, because ONNX apparently still has memory management issues)?