Build uploaded using `kernels`.

.gitattributes

@@ -51,3 +51,12 @@ build/torch28-cxx11-cu129-x86_64-linux/mra/_mra_9e0f4db.abi3.so filter=lfs diff=
 build/torch29-cxx11-cu126-x86_64-linux/mra/_mra_9e0f4db.abi3.so filter=lfs diff=lfs merge=lfs -text
 build/torch29-cxx11-cu128-x86_64-linux/mra/_mra_9e0f4db.abi3.so filter=lfs diff=lfs merge=lfs -text
 build/torch29-cxx11-cu130-x86_64-linux/mra/_mra_9e0f4db.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch210-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch210-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch210-cxx11-cu130-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch28-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch28-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch28-cxx11-cu129-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch29-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch29-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text
+build/torch29-cxx11-cu130-x86_64-linux/_mra_b91b835.abi3.so filter=lfs diff=lfs merge=lfs -text

build/torch210-cxx11-cu126-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/{torch28-cxx11-cu128-x86_64-linux/mra/_mra_9e0f4db.abi3.so → torch210-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dffa76035f51278977f311ed5acc0ad772b78ccec8dc376a672ca89e2f28d024
-size 2602848
+oid sha256:fb5be144697731fba4406b4ed16232dca9b2c05cb7715646efe3679e7b6343b0
+size 2451456

build/{torch28-cxx11-cu128-x86_64-linux/mra → torch210-cxx11-cu126-x86_64-linux}/_ops.py

@@ -1,9 +1,9 @@
 import torch
-from . import _mra_9e0f4db
-ops = torch.ops._mra_9e0f4db
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
 
 def add_op_namespace_prefix(op_name: str):
     """
     Prefix op by namespace.
     """
-    return f"_mra_9e0f4db::{op_name}"
+    return f"_mra_b91b835::{op_name}"

build/torch210-cxx11-cu126-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch210-cxx11-cu126-x86_64-linux/mra/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import sys
+
+import importlib
+from pathlib import Path
+from types import ModuleType
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch210-cxx11-cu128-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/{torch28-cxx11-cu129-x86_64-linux/mra/_mra_9e0f4db.abi3.so → torch210-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6f6fd347cc0a80dcca1709760e4ec6495fc2b6a7bf89070b1a37b772bc2d79a0
-size 2632536
+oid sha256:2719092133c64747d7527e90b18e4bf88616787205bc49ca513544b8d93b7db3
+size 2719824

build/{torch28-cxx11-cu129-x86_64-linux/mra → torch210-cxx11-cu128-x86_64-linux}/_ops.py

@@ -1,9 +1,9 @@
 import torch
-from . import _mra_9e0f4db
-ops = torch.ops._mra_9e0f4db
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
 
 def add_op_namespace_prefix(op_name: str):
     """
     Prefix op by namespace.
     """
-    return f"_mra_9e0f4db::{op_name}"
+    return f"_mra_b91b835::{op_name}"

build/torch210-cxx11-cu128-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch210-cxx11-cu128-x86_64-linux/mra/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import sys
+
+import importlib
+from pathlib import Path
+from types import ModuleType
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch210-cxx11-cu130-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/{torch28-cxx11-cu126-x86_64-linux/mra/_mra_9e0f4db.abi3.so → torch210-cxx11-cu130-x86_64-linux/_mra_b91b835.abi3.so}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16b83b6d735eb5a16132b29c3ee6594379a132c812ffbbc668bdbe7b28aa4d23
-size 2334488
+oid sha256:4e562370de6b1e766e9fa49f731fc4516ec726643cdbb0a57d6a291359ea5954
+size 2641336

build/{torch29-cxx11-cu126-x86_64-linux/mra → torch210-cxx11-cu130-x86_64-linux}/_ops.py

@@ -1,9 +1,9 @@
 import torch
-from . import _mra_9e0f4db
-ops = torch.ops._mra_9e0f4db
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
 
 def add_op_namespace_prefix(op_name: str):
     """
     Prefix op by namespace.
     """
-    return f"_mra_9e0f4db::{op_name}"
+    return f"_mra_b91b835::{op_name}"

build/torch210-cxx11-cu130-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch210-cxx11-cu130-x86_64-linux/mra/__init__.py

@@ -0,0 +1,26 @@
+import ctypes
+import sys
+
+import importlib
+from pathlib import Path
+from types import ModuleType
+
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
+
+
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch28-cxx11-cu126-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/{torch29-cxx11-cu126-x86_64-linux/mra/_mra_9e0f4db.abi3.so → torch28-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b54153235319a97ed3a32d8938b70823e3204bfff9503c7efe881228014d6339
-size 2334464
+oid sha256:66023954635a31836ddef4a3d072796477d2df17611a3b0acadc83e4b9d644ad
+size 2446064

build/torch28-cxx11-cu126-x86_64-linux/{mra/_ops.py → _ops.py}

@@ -1,9 +1,9 @@
 import torch
-from . import _mra_9e0f4db
-ops = torch.ops._mra_9e0f4db
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
 
 def add_op_namespace_prefix(op_name: str):
     """
     Prefix op by namespace.
     """
-    return f"_mra_9e0f4db::{op_name}"
+    return f"_mra_b91b835::{op_name}"

build/torch28-cxx11-cu126-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch28-cxx11-cu126-x86_64-linux/mra/__init__.py

@@ -1,25 +1,26 @@
-from ._ops import ops
-import torch
+import ctypes
+import sys
 
-def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+import importlib
+from pathlib import Path
+from types import ModuleType
 
-def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
-    return ops.mm_to_sparse(dense_A, dense_B, indices)
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
 
-def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
-    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
 
-def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
-
-def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
-    return ops.scatter(dense_A, indices, B_num_block)
-
-__all__ = [
-    "index_max",
-    "mm_to_sparse",
-    "sparse_dense_mm",
-    "reduce_sum",
-    "scatter",
-]
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch28-cxx11-cu128-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/torch28-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daaac316543e2c585eb477e8c494ee342f46e292048e6ac0d3637fce1577bf68
+size 2714440

build/torch28-cxx11-cu128-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_mra_b91b835::{op_name}"

build/torch28-cxx11-cu128-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch28-cxx11-cu128-x86_64-linux/mra/__init__.py

@@ -1,25 +1,26 @@
-from ._ops import ops
-import torch
+import ctypes
+import sys
 
-def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+import importlib
+from pathlib import Path
+from types import ModuleType
 
-def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
-    return ops.mm_to_sparse(dense_A, dense_B, indices)
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
 
-def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
-    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
 
-def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
-
-def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
-    return ops.scatter(dense_A, indices, B_num_block)
-
-__all__ = [
-    "index_max",
-    "mm_to_sparse",
-    "sparse_dense_mm",
-    "reduce_sum",
-    "scatter",
-]
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch28-cxx11-cu129-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/torch28-cxx11-cu129-x86_64-linux/_mra_b91b835.abi3.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9438df3b497af7e80e214b74bf4b9e789d3cd1f856105dc251c5de6ba750bdb4
+size 2748224

build/torch28-cxx11-cu129-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_mra_b91b835::{op_name}"

build/torch28-cxx11-cu129-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch28-cxx11-cu129-x86_64-linux/mra/__init__.py

@@ -1,25 +1,26 @@
-from ._ops import ops
-import torch
+import ctypes
+import sys
 
-def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+import importlib
+from pathlib import Path
+from types import ModuleType
 
-def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
-    return ops.mm_to_sparse(dense_A, dense_B, indices)
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
 
-def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
-    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
 
-def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
-
-def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
-    return ops.scatter(dense_A, indices, B_num_block)
-
-__all__ = [
-    "index_max",
-    "mm_to_sparse",
-    "sparse_dense_mm",
-    "reduce_sum",
-    "scatter",
-]
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch29-cxx11-cu126-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/torch29-cxx11-cu126-x86_64-linux/_mra_b91b835.abi3.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eafd95e7fa0334563e9b39d8381fa91af8f65d9d654185ffcf63a07e9be1b6e3
+size 2446040

build/torch29-cxx11-cu126-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_mra_b91b835::{op_name}"

build/torch29-cxx11-cu126-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch29-cxx11-cu126-x86_64-linux/mra/__init__.py

@@ -1,25 +1,26 @@
-from ._ops import ops
-import torch
+import ctypes
+import sys
 
-def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+import importlib
+from pathlib import Path
+from types import ModuleType
 
-def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
-    return ops.mm_to_sparse(dense_A, dense_B, indices)
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
 
-def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
-    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
 
-def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
-
-def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
-    return ops.scatter(dense_A, indices, B_num_block)
-
-__all__ = [
-    "index_max",
-    "mm_to_sparse",
-    "sparse_dense_mm",
-    "reduce_sum",
-    "scatter",
-]
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))

build/torch29-cxx11-cu128-x86_64-linux/__init__.py

@@ -0,0 +1,25 @@
+from ._ops import ops
+import torch
+
+def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+
+def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
+    return ops.mm_to_sparse(dense_A, dense_B, indices)
+
+def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
+    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
+
+def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
+    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
+
+def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
+    return ops.scatter(dense_A, indices, B_num_block)
+
+__all__ = [
+    "index_max",
+    "mm_to_sparse",
+    "sparse_dense_mm",
+    "reduce_sum",
+    "scatter",
+]

build/torch29-cxx11-cu128-x86_64-linux/_mra_b91b835.abi3.so

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3c5b1dc0d06c9c9bd4813ef6617a3d61d7fe8785be8b000c69a7ecf800572d6
+size 2714408

build/torch29-cxx11-cu128-x86_64-linux/_ops.py

@@ -0,0 +1,9 @@
+import torch
+from . import _mra_b91b835
+ops = torch.ops._mra_b91b835
+
+def add_op_namespace_prefix(op_name: str):
+    """
+    Prefix op by namespace.
+    """
+    return f"_mra_b91b835::{op_name}"

build/torch29-cxx11-cu128-x86_64-linux/metadata.json

@@ -0,0 +1 @@
+{"python-depends":[]}

build/torch29-cxx11-cu128-x86_64-linux/mra/__init__.py

@@ -1,25 +1,26 @@
-from ._ops import ops
-import torch
+import ctypes
+import sys
 
-def index_max(index_vals: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.index_max(index_vals, indices, A_num_block, B_num_block)
+import importlib
+from pathlib import Path
+from types import ModuleType
 
-def mm_to_sparse(dense_A: torch.Tensor, dense_B: torch.Tensor, indices: torch.Tensor):
-    return ops.mm_to_sparse(dense_A, dense_B, indices)
+def _import_from_path(file_path: Path) -> ModuleType:
+    # We cannot use the module name as-is, after adding it to `sys.modules`,
+    # it would also be used for other imports. So, we make a module name that
+    # depends on the path for it to be unique using the hex-encoded hash of
+    # the path.
+    path_hash = "{:x}".format(ctypes.c_size_t(hash(file_path.absolute())).value)
+    module_name = path_hash
+    spec = importlib.util.spec_from_file_location(module_name, file_path)
+    if spec is None:
+        raise ImportError(f"Cannot load spec for {module_name} from {file_path}")
+    module = importlib.util.module_from_spec(spec)
+    if module is None:
+        raise ImportError(f"Cannot load module {module_name} from spec")
+    sys.modules[module_name] = module
+    spec.loader.exec_module(module)  # type: ignore
+    return module
 
-def sparse_dense_mm(sparse_A: torch.Tensor, indices: torch.Tensor, dense_B: torch.Tensor, A_num_block: int):
-    return ops.sparse_dense_mm(sparse_A, indices, dense_B, A_num_block)
 
-def reduce_sum(sparse_A: torch.Tensor, indices: torch.Tensor, A_num_block: int, B_num_block: int):
-    return ops.reduce_sum(sparse_A, indices, A_num_block, B_num_block)
-
-def scatter(dense_A: torch.Tensor, indices: torch.Tensor, B_num_block: int):
-    return ops.scatter(dense_A, indices, B_num_block)
-
-__all__ = [
-    "index_max",
-    "mm_to_sparse",
-    "sparse_dense_mm",
-    "reduce_sum",
-    "scatter",
-]
+globals().update(vars(_import_from_path(Path(__file__).parent.parent / "__init__.py")))