Skip to content

Commit 5f12f28

Browse files
qmafacebook-github-bot
authored andcommitted
Enable quantization as default for XNNPack for previous failing models (pytorch#6242)
Summary: Pull Request resolved: pytorch#6242 Since master has migrated aot_compiler to use to_edge_transform_and_lower in a previous change pytorch#6026, quantization XNNPack options can be enabled by default for the following models: - Quantized ViT - Quantized Mobilebert - Quantized Emformer Predict - Quantized Emformer Transcribe Reviewed By: digantdesai Differential Revision: D64081319 fbshipit-source-id: 4e8ff77af442dfded043c5a5583466afec6beb4e
1 parent ab628cc commit 5f12f28

File tree

1 file changed

+6
-4
lines changed

1 file changed

+6
-4
lines changed

examples/xnnpack/__init__.py

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,8 @@
44
# This source code is licensed under the BSD-style license found in the
55
# LICENSE file in the root directory of this source tree.
66

7+
# pyre-unsafe
8+
79
from dataclasses import dataclass
810

911

@@ -24,14 +26,14 @@ class XNNPACKOptions(object):
2426
"mv3": XNNPACKOptions(True, True),
2527
"resnet18": XNNPACKOptions(True, True),
2628
"resnet50": XNNPACKOptions(True, True),
27-
"vit": XNNPACKOptions(False, True), # T161242362
29+
"vit": XNNPACKOptions(True, True),
2830
"w2l": XNNPACKOptions(True, True),
2931
"edsr": XNNPACKOptions(True, True),
30-
"mobilebert": XNNPACKOptions(False, True), # T197452682
32+
"mobilebert": XNNPACKOptions(True, True),
3133
"llama2": XNNPACKOptions(False, True),
3234
"emformer_join": XNNPACKOptions(True, True),
33-
"emformer_predict": XNNPACKOptions(False, True), # T197457838
34-
"emformer_transcribe": XNNPACKOptions(False, True), # T197449765
35+
"emformer_predict": XNNPACKOptions(True, True),
36+
"emformer_transcribe": XNNPACKOptions(True, True),
3537
}
3638

3739

0 commit comments

Comments
 (0)