If someone picks this up before me -
having looked at torch for similar ops (also in torchtext) and not finding a direct mapping (maybe some ops from torchtext can be used paritally)
I would suggest, to take the sklearn impl (which is widely used in practice) for TFIDFVectorization
Or
the onnx impl as reference (convert to pyTorch impl first) to add the onnx to torch lowering.
If someone picks this up before me - having looked at torch for similar ops (also in torchtext) and not finding a direct mapping (maybe some ops from torchtext can be used paritally) I would suggest, to take the sklearn impl (which is widely used in practice) for TFIDFVectorization Or the onnx impl as reference (convert to pyTorch impl first) to add the onnx to torch lowering.