Heads up: deprecating the convert {To,From} MLIRSparseTensor methods

The MLIR Sparsifier team is planning to deprecate these two library methods (which materialize sparse tensors from external buffers to COO format) in favor of much more flexible ops. The methods were for internal use only (e.g. MLIR’s PyTACO and the Python tests). However, if your project uses these methods, you may want to migrate to the new ops (contact us for details).

1 Like