Skip to content

Commit 98d7f55

Browse files
committed
Integrate Automated QDQ placement tool - part 3.2
Signed-off-by: Will Guo <willg@nvidia.com>
1 parent e53ca61 commit 98d7f55

3 files changed

Lines changed: 1499 additions & 0 deletions

File tree

Lines changed: 62 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,62 @@
1+
# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2+
# SPDX-License-Identifier: Apache-2.0
3+
#
4+
# Licensed under the Apache License, Version 2.0 (the "License");
5+
# you may not use this file except in compliance with the License.
6+
# You may obtain a copy of the License at
7+
#
8+
# http://www.apache.org/licenses/LICENSE-2.0
9+
#
10+
# Unless required by applicable law or agreed to in writing, software
11+
# distributed under the License is distributed on an "AS IS" BASIS,
12+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13+
# See the License for the specific language governing permissions and
14+
# limitations under the License.
15+
16+
"""Pattern-Based Q/DQ Autotuning for ONNX Models.
17+
18+
This package provides automated optimization of Quantize/Dequantize (Q/DQ) node placement
19+
in ONNX computation graphs to minimize TensorRT inference latency. It uses pattern-based
20+
region analysis to efficiently explore and optimize Q/DQ insertion strategies.
21+
"""
22+
23+
# Core data structures
24+
from .autotuner import QDQAutotuner
25+
from .common import (
26+
AutotunerError,
27+
AutotunerNotInitializedError,
28+
Config,
29+
InsertionScheme,
30+
InvalidSchemeError,
31+
PatternCache,
32+
PatternSchemes,
33+
Region,
34+
RegionType,
35+
)
36+
from .insertion_points import (
37+
ChildRegionInputInsertionPoint,
38+
ChildRegionOutputInsertionPoint,
39+
NodeInputInsertionPoint,
40+
ResolvedInsertionPoint,
41+
)
42+
from .region_pattern import RegionPattern
43+
from .region_search import CombinedRegionSearch
44+
45+
__all__ = [
46+
"AutotunerError",
47+
"AutotunerNotInitializedError",
48+
"ChildRegionInputInsertionPoint",
49+
"ChildRegionOutputInsertionPoint",
50+
"CombinedRegionSearch",
51+
"Config",
52+
"InsertionScheme",
53+
"InvalidSchemeError",
54+
"NodeInputInsertionPoint",
55+
"PatternCache",
56+
"PatternSchemes",
57+
"QDQAutotuner",
58+
"Region",
59+
"RegionPattern",
60+
"RegionType",
61+
"ResolvedInsertionPoint",
62+
]

0 commit comments

Comments
 (0)