diff --git a/python/tvm/topi/cuda/sparse.py b/python/tvm/topi/cuda/sparse.py index 3717c20e3a237..76952774c35e2 100644 --- a/python/tvm/topi/cuda/sparse.py +++ b/python/tvm/topi/cuda/sparse.py @@ -129,9 +129,9 @@ def _callback(op): s[op].bind(ni, thread_x) a, _ = s[c].split(s[c].op.axis[1], factor=1) s[c].bind(a, thread_x) - ao, ai = s[op].split(mi, nparts=4) + ao, _ = s[op].split(mi, nparts=4) s[op].bind(ao, thread_y) - ao, ai = s[c].split(s[c].op.axis[0], nparts=4) + ao, _ = s[c].split(s[c].op.axis[0], nparts=4) s[c].bind(ao, thread_y) traverse_inline(s, out.op, _callback) @@ -285,7 +285,7 @@ def gen_ir(data, w_data, w_indices, w_indptr, out): @autotvm.register_topi_compute("sparse_dense_padded.cuda") -def sparse_dense_padded(cfg, data, weight_data, weight_indices, weight_indptr): +def sparse_dense_padded(_, data, weight_data, weight_indices, weight_indptr): """ Computes sparse-dense matrix multiplication of `data` and `(weight_data, weight_indices, weight_indptr).T` @@ -321,7 +321,7 @@ def sparse_dense_padded(cfg, data, weight_data, weight_indices, weight_indptr): @autotvm.register_topi_schedule("sparse_dense_padded.cuda") -def schedule_sparse_dense_padded(cfg, outs): +def schedule_sparse_dense_padded(_, outs): """Create schedule for sparse dense""" # XXX: this will fail if we don't include the data_t Tensor in the schedule # ops. Maybe create_schedule should do some analysis so this isn't @@ -365,7 +365,7 @@ def pad_sparse_matrix(matrix, blocksize): @nn.sparse_dense_alter_layout.register(["cuda", "gpu"]) -def _alter_sparse_dense_layout(attrs, inputs, tinfos, out_type): +def _alter_sparse_dense_layout(_attrs, inputs, _tinfos, _out_type): """With cuda, we modify use alter_op_layout to swap the default sparse_dense implementation for one that operates on a padded matrix. We also padd the matrix. diff --git a/python/tvm/topi/nn/sparse.py b/python/tvm/topi/nn/sparse.py index 86a63b9659a96..74a9ad5fd6501 100644 --- a/python/tvm/topi/nn/sparse.py +++ b/python/tvm/topi/nn/sparse.py @@ -210,7 +210,7 @@ def _csr_transpose_ir(data, indices, indptr, out_data, out_indices, out_indptr): @tvm.target.generic_func -def sparse_dense_alter_layout(attrs, inputs, tinfos, out_type): +def sparse_dense_alter_layout(_attrs, _inputs, _tinfos, _out_type): """Change Sparse Dense layout. This is used for modifying the inputs weights so they are more amenable for