2 views

NVIDIA/Model-Optimizer

A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks like TensorRT-LLM, TensorRT, vLLM, etc. to optimize inference speed.

Python★ 4,807+357 that day#2 on trending

Analyzed September 27, 2026 at 23355ed (main) with jscpd 5.4.0 in 1.2s. Code duplication stayed at 1.85% across 3 appearances since Sep 25.

C66.3/100

Project health379,979 lines of code (L)

Duplication77

3.1% in python, bash, cpp, c-header, perl, javascript

Dead code78

2.6%

Complexity48

53%

The score is jscpd's own --health model: markup, data and text left out, tests kept. The numbers below are the trending scan's.

1.85%duplicated code · 3,538 of 191,625 lines

240clones in code

608code files of 2,070 scanned

4.16%duplicated over all files · 1,248 clones

Most complex files

Dead code 2.63% unused · 122 findings

modelopt/torch/sparsity/attention_sparsity/plugins/vllm.pyunused-file
tools/launcher/core.pyunused-file
modelopt/torch/utils/mlflow.pyunused-file
modelopt/torch/utils/plugins/transformers_dataset.pyunused-file
modelopt/torch/sparsity/attention_sparsity/plugins/vllm_runtime.pyunused-file
modelopt/torch/speculative/plugins/hf_streaming_dataset.pyunused-file
modelopt/torch/speculative/plugins/rdma_hidden_states_connector.pyunused-file
modelopt/torch/kernels/common/attention/decode_attention.pyunused-file
modelopt/torch/utils/plugins/mbridge.pyunused-file
modelopt/torch/sparsity/attention_sparsity/plugins/sparse_attn_calibration.pyunused-file

The code number leaves out tests, fixtures, docs, examples, data files, styles, translations and vendored code.

Largest clones in the code

18 lines in 2 placescpp684 tokens

modelopt/onnx/quantization/src/modelopt_round_and_pack_ext.cpp:1–18modelopt/torch/kernels/quantization/gemm/tensor_quant.cpp:1–18

1/*
2 * SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3 * SPDX-License-Identifier: Apache-2.0
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *

First 10 of 18 lines.

18 lines in 2 placesc-header683 tokens

modelopt/torch/kernels/quantization/gemm/tensor_quant.h:1–18modelopt/torch/kernels/quantization/gemm/tensor_quant_mx.h:1–18

1/*
2 * SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3 * SPDX-License-Identifier: Apache-2.0
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *

First 10 of 18 lines.

16 lines in 2 placescpp682 tokens

modelopt/onnx/quantization/src/modelopt_round_and_pack_ext.cpp:1–16modelopt/torch/kernels/quantization/conv/implicit_gemm_binding.cpp:1–16

1/*
2 * SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3 * SPDX-License-Identifier: Apache-2.0
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *

First 10 of 16 lines.

7 more

15 lines in 2 placescpp636 tokens
modelopt/onnx/quantization/src/modelopt_round_and_pack_ext.cpp:2–16 modelopt/torch/kernels/quantization/ggml/ggml.cpp:2–16
67 lines in 2 placespython496 tokens
modelopt/torch/_deploy/compilation.py:40–106 modelopt/torch/_deploy/profiling.py:35–102
66 lines in 2 placespython493 tokens
modelopt/torch/_deploy/compilation.py:40–105 modelopt/torch/_deploy/profiling.py:119–184
30 lines in 2 placespython390 tokens
modelopt/torch/nas/algorithms.py:525–554 modelopt/torch/prune/pruning.py:127–156
48 lines in 2 placespython364 tokens
modelopt/torch/_deploy/compilation.py:41–88 modelopt/torch/nas/algorithms.py:605–652
47 lines in 2 placespython363 tokens
modelopt/torch/_deploy/compilation.py:42–88 modelopt/torch/nas/algorithms.py:483–529
46 lines in 2 placespython333 tokens
modelopt/torch/_deploy/compilation.py:43–88 modelopt/torch/prune/pruning.py:86–131

Largest clones over all files, tests and data included

16 lines in 7 placescppdocs and examples682 tokens
examples/onnx_ptq/custom_op_plugin/plugin/IdentityConvPlugin.cpp:1–16 examples/onnx_ptq/custom_op_plugin/plugin/IdentityConvPluginCreator.cpp:1–16 examples/onnx_ptq/custom_op_plugin/plugin/PluginRegistration.cpp:1–16 examples/onnx_ptq/custom_op_plugin/plugin/PluginUtils.cpp:1–16 modelopt/onnx/quantization/src/modelopt_round_and_pack_ext.cpp:1–16and 2 more
```
1/*
2 * SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3 * SPDX-License-Identifier: Apache-2.0
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *
```
First 10 of 16 lines.
16 lines in 5 placesc-headerdocs and examples682 tokens
examples/onnx_ptq/custom_op_plugin/plugin/IdentityConvPlugin.h:1–16 examples/onnx_ptq/custom_op_plugin/plugin/IdentityConvPluginCreator.h:1–16 examples/onnx_ptq/custom_op_plugin/plugin/PluginUtils.h:1–16 modelopt/torch/kernels/quantization/gemm/tensor_quant.h:1–16 modelopt/torch/kernels/quantization/gemm/tensor_quant_mx.h:1–16
```
1/*
2 * SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
3 * SPDX-License-Identifier: Apache-2.0
4 *
5 * Licensed under the Apache License, Version 2.0 (the "License");
6 * you may not use this file except in compliance with the License.
7 * You may obtain a copy of the License at
8 *
9 * http://www.apache.org/licenses/LICENSE-2.0
10 *
```
First 10 of 16 lines.

6 lines in 7 placesmarkdowndocs and examples150 tokens

examples/diffusers/README.md:482–487examples/hf_ptq/README.md:792–797examples/llm_distill/README.md:182–187examples/onnx_ptq/README.md:273–278examples/speculative_decoding/README.md:350–355and 2 more

482- 📅 [Roadmap](https://github.com/NVIDIA/Model-Optimizer/issues/1699)
483- 📖 [Documentation](https://nvidia.github.io/Model-Optimizer)
484- 🎯 [Benchmarks](../benchmark.md)
485- 💡 [Release Notes](https://nvidia.github.io/Model-Optimizer/reference/0_changelog.html)
486- 🐛 [File a bug](https://github.com/NVIDIA/Model-Optimizer/issues/new?template=1_bug_report.md)
487- ✨ [File a Feature Request](https://github.com/NVIDIA/Model-Optimizer/issues/new?template=2_feature_request.md)

15 lines in 2 placescppdocs and examples636 tokens
examples/onnx_ptq/custom_op_plugin/plugin/IdentityConvPlugin.cpp:2–16 modelopt/torch/kernels/quantization/ggml/ggml.cpp:2–16
122 lines in 2 placespythondocs and examples577 tokens
examples/llm_eval/mmlu.py:62–183 examples/windows/accuracy_benchmark/mmlu_benchmark.py:46–167

By format, code only

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	558	185,988	229	3,383	1.82%
bash	43	4,809	5	73	1.52%
cpp	4	339	4	55	16.22%
c-header	2	315	2	27	8.57%
perl	1	174	0	0	0%

By format, all files

Format	Files	Lines	Clones	Duplicated lines	Duplication
python	1,247	371,685	870	11,619	3.13%
yaml	394	22,513	302	5,732	25.46%
markdown	187	27,856	45	433	1.55%
bash	159	10,662	17	176	1.65%
cpp	8	756	8	117	15.48%
c-header	5	519	5	73	14.07%
txt	10	140	1	9	6.43%
json	17	923	0	0	0%
text	26	572	0	0	0%
toml	10	566	0	0	0%

Trending appearances

Day	Rank	Stars	Files	Lines	Clones	All files	Code	Health	Commit
Sep 27, 2026	#2	4,807 (+357)	2,070	436,739	1,248	4.16%	1.85%	C 66.3	`23355ed`
Sep 26, 2026	#9	4,526 (+359)	2,070	436,739	1,248	4.16%	1.85%	C 66.3	`23355ed`
Sep 25, 2026	#3	4,148 (+44)	2,070	436,738	1,248	4.16%	1.85%	C 66.3	`ed7e879`

Want this for your own project? Install jscpd and run jscpd ., or measure it the way the baselines were measured.