tvm.sparse

Python-interface for Sparse-TIR

Program transformations

`lower_sparse_iter`(mod)	Lower sparse iterators in Sparse TIR.
`lower_sparse_buffer`(mod)	Lower sparse buffers in Sparse TIR.
`format_decompose`(mod, composable_formats[, ...])	Rewrite the sparse format of sparse buffers in the TIR scripts.
`specialize_buffer`(mod, buf_name, idx_map)	Specialize a buffer in an IRModule with given buffer name and index map function.

FormatRewriteRule(name, new_format_desc, ...)

Format rewriting rule.

`condense`(indptr_nd, indices_nd, t, g)	Condense sparse matrix in CSR format to (t x 1) tiles, and group g tiles together.
`column_part_hyb`(num_rows, num_cols, ...)	Partition input CSR matrix by columns and collect rows into buckets according to non zero elements per row.
`csf_to_ell3d`(csf_indptr_0, csf_indices_0, ...)	Convert CSF format to composable ELL format in 3-dimensional setting (HeteroGraphs).