Today, we are converting scalar values into 1D tensors with a size of 1. Perhaps we can consider supporting 0 rank tensors(tensor<f32>, not f32). This is a special case that needs to be handled when constructing and passing into the runtime, but it may not be a significant issue.
Some ops can produce scalar values, like Matmul vector-vector product: https://github.com/tenstorrent/tt-mlir/pull/745#discussion_r1764954874
Today, we are converting scalar values into 1D tensors with a size of 1. Perhaps we can consider supporting 0 rank tensors(
tensor<f32>
, notf32
). This is a special case that needs to be handled when constructing and passing into the runtime, but it may not be a significant issue.