Skip to content

Commit dcb5b77

Browse files
tqchentqchen
committed
Lint
1 parent 6a08dbd commit dcb5b77

File tree

6 files changed

+57
-59
lines changed

6 files changed

+57
-59
lines changed

‎python/tvm/ir/attrs.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -14,7 +14,7 @@
1414
# KIND, either express or implied. See the License for the
1515
# specific language governing permissions and limitations
1616
# under the License.
17-
""" TVM Attribute module, which is mainly used for defining attributes of operators."""
17+
"""TVM Attribute module, which is mainly used for defining attributes of operators."""
1818
import tvm._ffi
1919

2020
from tvm.runtime import Object

‎python/tvm/relax/expr.py‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -827,9 +827,11 @@ def __init__(
827827
)
828828

829829
self.__init_handle_by_constructor__(
830-
_ffi_api.DataflowVar # type: ignore
831-
if isinstance(name_hint, str)
832-
else _ffi_api.DataflowVarFromId, # type: ignore
830+
(
831+
_ffi_api.DataflowVar # type: ignore
832+
if isinstance(name_hint, str)
833+
else _ffi_api.DataflowVarFromId
834+
), # type: ignore
833835
name_hint,
834836
struct_info,
835837
span,

‎python/tvm/relax/frontend/onnx/onnx_frontend.py‎

Lines changed: 2 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -2906,15 +2906,11 @@ def _impl_v11(cls, bb, inputs, attr, params):
29062906
mode = attr.get("mode", b"DCR").decode("utf-8")
29072907
b, c, h, w = inputs[0].struct_info.shape
29082908
if mode == "DCR":
2909-
x = relax.op.reshape(
2910-
inputs[0], (b, block_size, block_size, c // (block_size**2), h, w)
2911-
)
2909+
x = relax.op.reshape(inputs[0], (b, block_size, block_size, c // (block_size**2), h, w))
29122910
x = relax.op.permute_dims(x, [0, 3, 4, 1, 5, 2])
29132911
return relax.op.reshape(x, (b, c // (block_size**2), h * block_size, w * block_size))
29142912
elif mode == "CRD":
2915-
x = relax.op.reshape(
2916-
inputs[0], (b, c // (block_size**2), block_size, block_size, h, w)
2917-
)
2913+
x = relax.op.reshape(inputs[0], (b, c // (block_size**2), block_size, block_size, h, w))
29182914
x = relax.op.permute_dims(x, [0, 1, 4, 2, 5, 3])
29192915
return relax.op.reshape(x, (b, c // (block_size**2), h * block_size, w * block_size))
29202916
else:

‎python/tvm/tir/analysis/analysis.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -165,7 +165,7 @@ def get_block_read_write_region(
165165

166166

167167
def calculate_allocated_bytes(
168-
func_or_mod: Union[PrimFunc, IRModule]
168+
func_or_mod: Union[PrimFunc, IRModule],
169169
) -> Union[Dict[str, int], Dict[str, Dict[str, int]]]:
170170
"""Calculate allocated memory per memory scope required by TIR PrimFuncs.
171171

‎tests/python/contrib/test_hexagon/infrastructure.py‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -16,7 +16,7 @@
1616
# under the License.
1717
# pylint: disable=invalid-name
1818

19-
""" Hexagon testing infrastructure """
19+
"""Hexagon testing infrastructure"""
2020

2121
import numpy
2222
import tvm

‎tests/python/relax/test_transform_merge_composite_functions.py‎

Lines changed: 47 additions & 47 deletions
Original file line numberDiff line numberDiff line change
@@ -33,12 +33,12 @@ def main(
3333
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
3434
cls = Conv2dReLUx2
3535
with R.dataflow():
36-
lv: R.Tensor(
37-
(1, 64, 56, 56), dtype="float32"
38-
) = cls.fused_relax_nn_conv2d_relax_nn_relu(data, weight1)
39-
gv: R.Tensor(
40-
(1, 64, 54, 54), dtype="float32"
41-
) = cls.fused_relax_nn_conv2d_relax_nn_relu1(lv, weight2)
36+
lv: R.Tensor((1, 64, 56, 56), dtype="float32") = (
37+
cls.fused_relax_nn_conv2d_relax_nn_relu(data, weight1)
38+
)
39+
gv: R.Tensor((1, 64, 54, 54), dtype="float32") = (
40+
cls.fused_relax_nn_conv2d_relax_nn_relu1(lv, weight2)
41+
)
4242
R.output(gv)
4343
return gv
4444

@@ -85,10 +85,10 @@ def main(
8585
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
8686
cls = Conv2dReLUx2_merged
8787
with R.dataflow():
88-
gv: R.Tensor(
89-
(1, 64, 54, 54), dtype="float32"
90-
) = cls.fused_relax_nn_conv2d_relax_nn_relu_relax_nn_conv2d_relax_nn_relu1_dnnl(
91-
data, weight1, weight2
88+
gv: R.Tensor((1, 64, 54, 54), dtype="float32") = (
89+
cls.fused_relax_nn_conv2d_relax_nn_relu_relax_nn_conv2d_relax_nn_relu1_dnnl(
90+
data, weight1, weight2
91+
)
9292
)
9393
R.output(gv)
9494
return gv
@@ -159,7 +159,7 @@ def main(
159159

160160
@R.function(private=True)
161161
def fused_relax_nn_gelu(
162-
lv: R.Tensor((1, 64, 54, 54), dtype="float32")
162+
lv: R.Tensor((1, 64, 54, 54), dtype="float32"),
163163
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
164164
R.func_attr({"Primitive": 1, "Composite": "compiler_A.gelu"})
165165
with R.dataflow():
@@ -169,7 +169,7 @@ def fused_relax_nn_gelu(
169169

170170
@R.function(private=True)
171171
def fused_relax_nn_relu(
172-
lv1: R.Tensor((1, 64, 54, 54), dtype="float32")
172+
lv1: R.Tensor((1, 64, 54, 54), dtype="float32"),
173173
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
174174
R.func_attr({"Primitive": 1, "Composite": "compiler_A.relu"})
175175
with R.dataflow():
@@ -243,7 +243,7 @@ def lv(
243243

244244
@R.function
245245
def lv1(
246-
lv11: R.Tensor((1, 64, 54, 54), dtype="float32")
246+
lv11: R.Tensor((1, 64, 54, 54), dtype="float32"),
247247
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
248248
# function attr dict
249249
R.func_attr({"Composite": "compiler_A.relu"})
@@ -257,7 +257,7 @@ def lv1(
257257

258258
@R.function
259259
def lv21(
260-
lv4: R.Tensor((1, 64, 54, 54), dtype="float32")
260+
lv4: R.Tensor((1, 64, 54, 54), dtype="float32"),
261261
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
262262
# function attr dict
263263
R.func_attr({"Composite": "compiler_A.gelu"})
@@ -292,10 +292,10 @@ def main(
292292
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
293293
cls = Diamond_merged
294294
with R.dataflow():
295-
gv5: R.Tensor(
296-
(1, 64, 54, 54), dtype="float32"
297-
) = cls.fused_relax_nn_conv2d_relax_nn_relu_relax_nn_gelu_relax_add_compiler_A(
298-
data2, weight2
295+
gv5: R.Tensor((1, 64, 54, 54), dtype="float32") = (
296+
cls.fused_relax_nn_conv2d_relax_nn_relu_relax_nn_gelu_relax_add_compiler_A(
297+
data2, weight2
298+
)
299299
)
300300
R.output(gv5)
301301
return gv5
@@ -321,7 +321,7 @@ def main(
321321

322322
@R.function(private=True)
323323
def fused_relax_nn_gelu(
324-
lv: R.Tensor((1, 64, 54, 54), dtype="float32")
324+
lv: R.Tensor((1, 64, 54, 54), dtype="float32"),
325325
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
326326
R.func_attr({"Primitive": 1, "Composite": "compiler_B.gelu"})
327327
with R.dataflow():
@@ -331,7 +331,7 @@ def fused_relax_nn_gelu(
331331

332332
@R.function(private=True)
333333
def fused_relax_nn_relu(
334-
lv1: R.Tensor((1, 64, 54, 54), dtype="float32")
334+
lv1: R.Tensor((1, 64, 54, 54), dtype="float32"),
335335
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
336336
R.func_attr({"Primitive": 1, "Composite": "compiler_A.relu"})
337337
with R.dataflow():
@@ -418,7 +418,7 @@ def lv(
418418

419419
@R.function
420420
def lv1(
421-
lv11: R.Tensor((1, 64, 54, 54), dtype="float32")
421+
lv11: R.Tensor((1, 64, 54, 54), dtype="float32"),
422422
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
423423
R.func_attr({"Composite": "compiler_A.relu"})
424424
with R.dataflow():
@@ -432,13 +432,13 @@ def lv1(
432432

433433
@R.function
434434
def fused_relax_nn_gelu1_compiler_B(
435-
lv2: R.Tensor((1, 64, 54, 54), dtype="float32")
435+
lv2: R.Tensor((1, 64, 54, 54), dtype="float32"),
436436
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
437437
R.func_attr({"Codegen": "compiler_B"})
438438

439439
@R.function
440440
def lv21(
441-
lv3: R.Tensor((1, 64, 54, 54), dtype="float32")
441+
lv3: R.Tensor((1, 64, 54, 54), dtype="float32"),
442442
) -> R.Tensor((1, 64, 54, 54), dtype="float32"):
443443
R.func_attr({"Composite": "compiler_B.gelu"})
444444
with R.dataflow():
@@ -489,7 +489,7 @@ def main(
489489

490490
@R.function(private=True)
491491
def fused_relax_nn_relu(
492-
x11: R.Tensor((10,), dtype="float32")
492+
x11: R.Tensor((10,), dtype="float32"),
493493
) -> R.Tensor((10,), dtype="float32"):
494494
R.func_attr({"Primitive": 1, "Composite": "compiler_A.relu"})
495495
with R.dataflow():
@@ -499,7 +499,7 @@ def fused_relax_nn_relu(
499499

500500
@R.function(private=True)
501501
def fused_relax_nn_gelu(
502-
x21: R.Tensor((10,), dtype="float32")
502+
x21: R.Tensor((10,), dtype="float32"),
503503
) -> R.Tensor((10,), dtype="float32"):
504504
R.func_attr({"Primitive": 1, "Composite": "compiler_A.gelu"})
505505
with R.dataflow():
@@ -575,10 +575,10 @@ def main(
575575
) -> R.Tensor((10,), dtype="float32"):
576576
cls = MultipleProducers_merged
577577
with R.dataflow():
578-
gv4: R.Tensor(
579-
(10,), dtype="float32"
580-
) = cls.fused_relax_nn_relu_relax_nn_gelu_relax_nn_relu_relax_nn_gelu_relax_add_compiler_A(
581-
x12, x22
578+
gv4: R.Tensor((10,), dtype="float32") = (
579+
cls.fused_relax_nn_relu_relax_nn_gelu_relax_nn_relu_relax_nn_gelu_relax_add_compiler_A(
580+
x12, x22
581+
)
582582
)
583583
R.output(gv4)
584584
return gv4
@@ -599,7 +599,7 @@ def main(x1: R.Tensor((10,), dtype="float32")) -> R.Tensor((10,), dtype="float32
599599

600600
@R.function(private=True)
601601
def fused_relax_nn_relu(
602-
x11: R.Tensor((10,), dtype="float32")
602+
x11: R.Tensor((10,), dtype="float32"),
603603
) -> R.Tensor((10,), dtype="float32"):
604604
R.func_attr({"Primitive": 1, "Composite": "compiler_A.relu"})
605605
with R.dataflow():
@@ -609,7 +609,7 @@ def fused_relax_nn_relu(
609609

610610
@R.function(private=True)
611611
def fused_relax_nn_gelu(
612-
x21: R.Tensor((10,), dtype="float32")
612+
x21: R.Tensor((10,), dtype="float32"),
613613
) -> R.Tensor((10,), dtype="float32"):
614614
R.func_attr({"Primitive": 1, "Composite": "compiler_A.gelu"})
615615
with R.dataflow():
@@ -644,7 +644,7 @@ def main(x1: R.Tensor((10,), dtype="float32")) -> R.Tensor((10,), dtype="float32
644644

645645
@R.function
646646
def fused_relax_nn_relu1_compiler_A(
647-
x11: R.Tensor((10,), dtype="float32")
647+
x11: R.Tensor((10,), dtype="float32"),
648648
) -> R.Tensor((10,), dtype="float32"):
649649
# function attr dict
650650
R.func_attr({"Codegen": "compiler_A"})
@@ -722,7 +722,7 @@ def main(
722722

723723
@R.function(private=True)
724724
def fused_relax_nn_relu(
725-
add2: R.Tensor((10,), dtype="float32")
725+
add2: R.Tensor((10,), dtype="float32"),
726726
) -> R.Tensor((10,), dtype="float32"):
727727
R.func_attr({"Primitive": 1, "Composite": "compiler_A.relu"})
728728
with R.dataflow():
@@ -742,7 +742,7 @@ def fused_relax_add(
742742

743743
@R.function(private=True)
744744
def fused_relax_nn_gelu(
745-
x31: R.Tensor((10,), dtype="float32")
745+
x31: R.Tensor((10,), dtype="float32"),
746746
) -> R.Tensor((10,), dtype="float32"):
747747
R.func_attr({"Primitive": 1, "Composite": "compiler_B.gelu"})
748748
with R.dataflow():
@@ -817,7 +817,7 @@ def lv31(add2: R.Tensor((10,), dtype="float32")) -> R.Tensor((10,), dtype="float
817817

818818
@R.function
819819
def fused_relax_nn_gelu1_compiler_B(
820-
x3: R.Tensor((10,), dtype="float32")
820+
x3: R.Tensor((10,), dtype="float32"),
821821
) -> R.Tensor((10,), dtype="float32"):
822822
R.func_attr({"Codegen": "compiler_B"})
823823

@@ -841,9 +841,9 @@ def main(
841841
cls = MergeCompilerRegionsExampleRef
842842
with R.dataflow():
843843
lv5: R.Tensor((10,), dtype="float32") = cls.fused_relax_nn_gelu1_compiler_B(x32)
844-
lv13: R.Tuple(
845-
R.Tensor((10,), dtype="float32"), R.Tensor((10,), dtype="float32")
846-
) = cls.fused_relax_add_relax_add_relax_nn_relu_compiler_A(x12, x22, lv5)
844+
lv13: R.Tuple(R.Tensor((10,), dtype="float32"), R.Tensor((10,), dtype="float32")) = (
845+
cls.fused_relax_add_relax_add_relax_nn_relu_compiler_A(x12, x22, lv5)
846+
)
847847
lv23: R.Tensor((10,), dtype="float32") = lv13[0]
848848
lv32: R.Tensor((10,), dtype="float32") = lv13[1]
849849
lv41: R.Tensor((10,), dtype="float32") = cls.fused_relax_nn_gelu1_compiler_B(lv23)
@@ -1097,9 +1097,9 @@ def main(
10971097
lv1: R.Tensor((784, 512), dtype="float32") = R.permute_dims(
10981098
linear_relu_stack_0_weight, axes=None
10991099
)
1100-
gv: R.Tensor(
1101-
(1, 512), dtype="float32"
1102-
) = cls.fused_relax_reshape_relax_matmul_tensorrt(inp_0, lv1)
1100+
gv: R.Tensor((1, 512), dtype="float32") = (
1101+
cls.fused_relax_reshape_relax_matmul_tensorrt(inp_0, lv1)
1102+
)
11031103
R.output(gv)
11041104
return gv
11051105

@@ -1130,7 +1130,7 @@ def main(A: R.Tensor([10], dtype="float32")) -> R.Tensor([10], dtype="float32"):
11301130

11311131
@R.function(private=True)
11321132
def fused_relax_nn_relu(
1133-
Input: R.Tensor([10], dtype="float32")
1133+
Input: R.Tensor([10], dtype="float32"),
11341134
) -> R.Tensor([10], dtype="float32"):
11351135
R.func_attr({"Composite": "compiler_A.relu", "Primitive": 1})
11361136
with R.dataflow():
@@ -1151,7 +1151,7 @@ def relu(
11511151

11521152
@R.function(private=True)
11531153
def fused_relax_nn_gelu(
1154-
Input: R.Tensor([10], dtype="float32")
1154+
Input: R.Tensor([10], dtype="float32"),
11551155
) -> R.Tensor([10], dtype="float32"):
11561156
R.func_attr({"Composite": "compiler_A.gelu", "Primitive": 1})
11571157
with R.dataflow():
@@ -1173,13 +1173,13 @@ def main(A: R.Tensor([10], dtype="float32")) -> R.Tensor([10], dtype="float32"):
11731173

11741174
@R.function
11751175
def fused_relax_nn_relu1_compiler_A(
1176-
Input: R.Tensor([10], dtype="float32")
1176+
Input: R.Tensor([10], dtype="float32"),
11771177
) -> R.Tensor([10], dtype="float32"):
11781178
R.func_attr({"Codegen": "compiler_A"})
11791179

11801180
@R.function
11811181
def composite_lambda(
1182-
Input: R.Tensor([10], dtype="float32")
1182+
Input: R.Tensor([10], dtype="float32"),
11831183
) -> R.Tensor([10], dtype="float32"):
11841184
R.func_attr({"Composite": "compiler_A.relu"})
11851185
with R.dataflow():
@@ -1203,13 +1203,13 @@ def relu(
12031203

12041204
@R.function
12051205
def fused_relax_nn_gelu1_compiler_A(
1206-
Input: R.Tensor([10], dtype="float32")
1206+
Input: R.Tensor([10], dtype="float32"),
12071207
) -> R.Tensor([10], dtype="float32"):
12081208
R.func_attr({"Codegen": "compiler_A"})
12091209

12101210
@R.function
12111211
def composite_lambda(
1212-
Input: R.Tensor([10], dtype="float32")
1212+
Input: R.Tensor([10], dtype="float32"),
12131213
) -> R.Tensor([10], dtype="float32"):
12141214
R.func_attr({"Composite": "compiler_A.gelu"})
12151215
with R.dataflow():

0 commit comments

Comments
 (0)