mct-nightly 2.2.0.20241127.529__py3-none-any.whl → 2.2.0.20241128.546__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (6) hide show
  1. {mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/METADATA +25 -34
  2. {mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/RECORD +6 -6
  3. model_compression_toolkit/__init__.py +1 -1
  4. {mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/LICENSE.md +0 -0
  5. {mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/WHEEL +0 -0
  6. {mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/top_level.txt +0 -0
{mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: mct-nightly
3
- Version: 2.2.0.20241127.529
3
+ Version: 2.2.0.20241128.546
4
4
  Summary: A Model Compression Toolkit for neural networks
5
5
  Home-page: UNKNOWN
6
6
  License: UNKNOWN
@@ -77,9 +77,9 @@ MCT supports various quantization methods as appears below.
77
77
 
78
78
  Quantization Method | Complexity | Computational Cost | API | Tutorial
79
79
  -------------------- | -----------|--------------------|---------|--------
80
- PTQ (Post Training Quantization) | Low | Low (~1-10 CPU minutes) | [PyTorch API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/pytorch_post_training_quantization.html) / [Keras API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/keras_post_training_quantization.html) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_post_training_quantization.ipynb"><img src="https://img.shields.io/badge/Pytorch-green"/></a> <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_post-training_quantization.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
81
- GPTQ (parameters fine-tuning using gradients) | Moderate | Moderate (~1-3 GPU hours) | [PyTorch API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/pytorch_gradient_post_training_quantization.html) / [Keras API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/keras_gradient_post_training_quantization.html) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_mobilenet_gptq.ipynb"><img src="https://img.shields.io/badge/PyTorch-green"/></a> <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_mobilenet_gptq.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
82
- QAT (Quantization Aware Training) | High | High (~12-36 GPU hours) | [QAT API](https://sony.github.io/model_optimization/docs/api/api_docs/index.html#qat) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_qat.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
80
+ PTQ (Post Training Quantization) | Low | Low (~1-10 CPU minutes) | [PyTorch API](https://sony.github.io/model_optimization/api/api_docs/methods/pytorch_post_training_quantization.html) / [Keras API](https://sony.github.io/model_optimization/api/api_docs/methods/keras_post_training_quantization.html) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_post_training_quantization.ipynb"><img src="https://img.shields.io/badge/Pytorch-green"/></a> <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_post-training_quantization.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
81
+ GPTQ (parameters fine-tuning using gradients) | Moderate | Moderate (~1-3 GPU hours) | [PyTorch API](https://sony.github.io/model_optimization/api/api_docs/methods/pytorch_gradient_post_training_quantization.html) / [Keras API](https://sony.github.io/model_optimization/api/api_docs/methods/keras_gradient_post_training_quantization.html) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_mobilenet_gptq.ipynb"><img src="https://img.shields.io/badge/PyTorch-green"/></a> <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_mobilenet_gptq.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
82
+ QAT (Quantization Aware Training) | High | High (~12-36 GPU hours) | [QAT API](https://sony.github.io/model_optimization/api/api_docs/index.html#qat) | <a href="https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_qat.ipynb"><img src="https://img.shields.io/badge/Keras-green"/></a>
83
83
 
84
84
  </p>
85
85
  </div>
@@ -87,9 +87,9 @@ QAT (Quantization Aware Training) | High | High (~12-36 GPU hours) | [QAT API](
87
87
  For each flow, **Quantization core** utilizes various algorithms and hyper-parameters for optimal [hardware-aware](https://github.com/sony/model_optimization/blob/main/model_compression_toolkit/target_platform_capabilities/README.md) quantization results.
88
88
  For further details, please see [Supported features and algorithms](#high-level-features-and-techniques).
89
89
 
90
- Required input:
91
- - Floating point model - 32bit model in either .pt or .keras format
92
- - Representative dataset - can be either provided by the user, or generated utilizing the [Data Generation](#data-generation-) capability
90
+ **Required input**: Floating point model - 32bit model in either .pt or .keras format
91
+
92
+ **Optional input**: Representative dataset - can be either provided by the user, or generated utilizing the [Data Generation](#data-generation-) capability
93
93
 
94
94
  <div align="center">
95
95
  <p align="center">
@@ -122,13 +122,13 @@ Generates synthetic images based on the statistics stored in the model's batch n
122
122
  The specifications of the method are detailed in the paper: _"**Data Generation for Hardware-Friendly Post-Training Quantization**"_ [5].
123
123
  __________________________________________________________________________________________________________
124
124
  ### Structured Pruning [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_pruning_mnist.ipynb)
125
- Reduces model size/complexity and ensures better channels utilization by removing redundant input channels from layers and reconstruction of layer weights. Read more ([Pytorch API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/pytorch_pruning_experimental.html) / [Keras API](https://sony.github.io/model_optimization/docs/api/api_docs/methods/keras_pruning_experimental.html)).
125
+ Reduces model size/complexity and ensures better channels utilization by removing redundant input channels from layers and reconstruction of layer weights. Read more ([Pytorch API](https://sony.github.io/model_optimization/api/api_docs/methods/pytorch_pruning_experimental.html) / [Keras API](https://sony.github.io/model_optimization/api/api_docs/methods/keras_pruning_experimental.html)).
126
126
  __________________________________________________________________________________________________________
127
127
  ### **Debugging and Visualization**
128
128
  **🎛️ Network Editor (Modify Quantization Configurations)** [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/keras/example_keras_network_editor.ipynb).
129
- Modify your model's quantization configuration for specific layers or apply a custom edit rule (e.g adjust layer's bit-width) using MCT’s network editor
129
+ Modify your model's quantization configuration for specific layers or apply a custom edit rule (e.g adjust layer's bit-width) using MCT’s network editor.
130
130
 
131
- **🖥️ Visualization**. Observe useful information for troubleshooting the quantized model's performance using TensorBoard. [Read more](https://sony.github.io/model_optimization/docs/guidelines/visualization.html).
131
+ **🖥️ Visualization**. Observe useful information for troubleshooting the quantized model's performance using TensorBoard. [Read more](https://sony.github.io/model_optimization/guidelines/visualization.html).
132
132
 
133
133
  **🔑 XQuant (Explainable Quantization)** [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/sony/model_optimization/blob/main/tutorials/notebooks/mct_features_notebooks/pytorch/example_pytorch_xquant.ipynb). Get valuable insights regarding the quality and success of the quantization process of your model. The report includes histograms and similarity metrics between the original float model and the quantized model in key points of the model. The report can be visualized using TensorBoard.
134
134
  __________________________________________________________________________________________________________
@@ -138,15 +138,15 @@ The specifications of the algorithm are detailed in the paper: _"**EPTQ: Enhance
138
138
  More details on how to use EPTQ via MCT can be found in the [GPTQ guidelines](https://github.com/sony/model_optimization/blob/main/model_compression_toolkit/gptq/README.md).
139
139
 
140
140
  ## <div align="center">Resources</div>
141
- * [User Guide](https://sony.github.io/model_optimization/docs/index.html) contains detailed information about MCT and guides you from installation through optimizing models for your edge AI applications.
141
+ * [User Guide](https://sony.github.io/model_optimization/index.html) contains detailed information about MCT and guides you from installation through optimizing models for your edge AI applications.
142
142
 
143
- * MCT's [API Docs](https://sony.github.io/model_optimization/docs/api/api_docs/) is seperated per quantization methods:
143
+ * MCT's [API Docs](https://sony.github.io/model_optimization/api/api_docs/) is separated per quantization methods:
144
144
 
145
- * [Post-training quantization](https://sony.github.io/model_optimization/docs/api/api_docs/index.html#ptq) | PTQ API docs
146
- * [Gradient-based post-training quantization](https://sony.github.io/model_optimization/docs/api/api_docs/index.html#gptq) | GPTQ API docs
147
- * [Quantization-aware training](https://sony.github.io/model_optimization/docs/api/api_docs/index.html#qat) | QAT API docs
145
+ * [Post-training quantization](https://sony.github.io/model_optimization/api/api_docs/index.html#ptq) | PTQ API docs
146
+ * [Gradient-based post-training quantization](https://sony.github.io/model_optimization/api/api_docs/index.html#gptq) | GPTQ API docs
147
+ * [Quantization-aware training](https://sony.github.io/model_optimization/api/api_docs/index.html#qat) | QAT API docs
148
148
 
149
- * [Debug](https://sony.github.io/model_optimization/docs/guidelines/visualization.html) – modify optimization process or generate explainable report
149
+ * [Debug](https://sony.github.io/model_optimization/guidelines/visualization.html) – modify optimization process or generate an explainable report
150
150
 
151
151
  * [Release notes](https://github.com/sony/model_optimization/releases)
152
152
 
@@ -180,25 +180,15 @@ Currently, MCT is being tested on various Python, Pytorch and TensorFlow version
180
180
  <img src="/docsrc/images/PoseEst.png" width="200">
181
181
  <img src="/docsrc/images/ObjDet.png" width="200">
182
182
 
183
- ### Pytorch
184
- We quantized classification networks from the torchvision library.
185
- In the following table we present the ImageNet validation results for these models:
186
-
187
- | Network Name | Float Accuracy | 8Bit Accuracy | Data-Free 8Bit Accuracy |
188
- |---------------------------|-----------------|-----------------|-------------------------|
189
- | MobileNet V2 [3] | 71.886 | 71.444 |71.29|
190
- | ResNet-18 [3] | 69.86 | 69.63 |69.53|
191
- | SqueezeNet 1.1 [3] | 58.128 | 57.678 ||
192
-
193
- ### Keras
194
183
  MCT can quantize an existing 32-bit floating-point model to an 8-bit fixed-point (or less) model without compromising accuracy.
195
- Below is a graph of [MobileNetV2](https://keras.io/api/applications/mobilenet/) accuracy on ImageNet vs average bit-width of weights (X-axis), using
196
- single-precision quantization, mixed-precision quantization, and mixed-precision quantization with GPTQ.
184
+ Below is a graph of [MobileNetV2](https://pytorch.org/vision/main/models/generated/torchvision.models.mobilenet_v2.html) accuracy on ImageNet vs average bit-width of weights (X-axis), using **single-precision** quantization, **mixed-precision** quantization, and mixed-precision quantization with GPTQ.
197
185
 
198
- <img src="https://github.com/sony/model_optimization/raw/main/docsrc/images/mbv2_accuracy_graph.png">
186
+ <p align="center">
187
+ <img src="/docsrc/images/torch_mobilenetv2.png" width="800">
199
188
 
200
189
  For more results, please see [1]
201
190
 
191
+
202
192
  ### Pruning Results
203
193
 
204
194
  Results for applying pruning to reduce the parameters of the following models by 50%:
@@ -210,19 +200,20 @@ Results for applying pruning to reduce the parameters of the following models by
210
200
 
211
201
  ## <div align="center">Troubleshooting and Community</div>
212
202
 
213
- If you encountered large accuracy degradation with MCT, check out the [Quantization Troubleshooting](https://github.com/sony/model_optimization/tree/main/quantization_troubleshooting.md)
214
- for common pitfalls and some tools to improve quantized model's accuracy.
203
+ If you encountered a large accuracy degradation with MCT, check out the [Quantization Troubleshooting](https://github.com/sony/model_optimization/tree/main/quantization_troubleshooting.md)
204
+ for common pitfalls and some tools to improve the quantized model's accuracy.
215
205
 
216
206
  Check out the [FAQ](https://github.com/sony/model_optimization/tree/main/FAQ.md) for common issues.
217
207
 
218
- You are welcome to ask questions and get support on our [issues section](https://github.com/sony/model_optimization/issues) and manage community discussions under [discussions section](https://github.com/sony/model_optimization/discussions).
208
+ You are welcome to ask questions and get support on our [issues section](https://github.com/sony/model_optimization/issues) and manage community discussions under the [discussions section](https://github.com/sony/model_optimization/discussions).
219
209
 
220
210
 
221
211
  ## <div align="center">Contributions</div>
222
- MCT aims at keeping a more up-to-date fork and welcomes contributions from anyone.
212
+ We'd love your input! MCT would not be possible without help from our community, and welcomes contributions from anyone!
223
213
 
224
214
  *Checkout our [Contribution guide](https://github.com/sony/model_optimization/blob/main/CONTRIBUTING.md) for more details.
225
215
 
216
+ Thank you 🙏 to all our contributors!
226
217
 
227
218
  ## <div align="center">License</div>
228
219
  MCT is licensed under Apache License Version 2.0. By contributing to the project, you agree to the license and copyright terms therein and release your contribution under these terms.
{mct_nightly-2.2.0.20241127.529.dist-info → mct_nightly-2.2.0.20241128.546.dist-info}/RECORD RENAMED
@@ -1,4 +1,4 @@
1
- model_compression_toolkit/__init__.py,sha256=TA9ze23Sb1Oe07ep1m1YS_g_HK4xeEFIaYjw126kAwI,1573
1
+ model_compression_toolkit/__init__.py,sha256=zVRBw5AaiemU9kcIUsX-NlE27jnB2iW9beBf0n-WvFA,1573
2
2
  model_compression_toolkit/constants.py,sha256=i4wYheBkIdQmsQA-axIpcT3YiSO1USNc-jaNiNE8w6E,3920
3
3
  model_compression_toolkit/defaultdict.py,sha256=LSc-sbZYXENMCw3U9F4GiXuv67IKpdn0Qm7Fr11jy-4,2277
4
4
  model_compression_toolkit/logger.py,sha256=3DByV41XHRR3kLTJNbpaMmikL8icd9e1N-nkQAY9oDk,4567
@@ -559,8 +559,8 @@ model_compression_toolkit/xquant/pytorch/model_analyzer.py,sha256=b93o800yVB3Z-i
559
559
  model_compression_toolkit/xquant/pytorch/pytorch_report_utils.py,sha256=bOc-hFL3gdoSM1Th_S2N_-9JJSlPGpZCTx_QLJHS6lg,3388
560
560
  model_compression_toolkit/xquant/pytorch/similarity_functions.py,sha256=CERxq5K8rqaiE-DlwhZBTUd9x69dtYJlkHOPLB54vm8,2354
561
561
  model_compression_toolkit/xquant/pytorch/tensorboard_utils.py,sha256=mkoEktLFFHtEKzzFRn_jCnxjhJolK12TZ5AQeDHzUO8,9767
562
- mct_nightly-2.2.0.20241127.529.dist-info/LICENSE.md,sha256=aYSSIb-5AFPeITTvXm1UAoe0uYBiMmSS8flvXaaFUks,10174
563
- mct_nightly-2.2.0.20241127.529.dist-info/METADATA,sha256=_BFXYbvbuoo6a7pXxnRT7YOIZO37P_0RyiGn-NXpMXA,26947
564
- mct_nightly-2.2.0.20241127.529.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
565
- mct_nightly-2.2.0.20241127.529.dist-info/top_level.txt,sha256=gsYA8juk0Z-ZmQRKULkb3JLGdOdz8jW_cMRjisn9ga4,26
566
- mct_nightly-2.2.0.20241127.529.dist-info/RECORD,,
562
+ mct_nightly-2.2.0.20241128.546.dist-info/LICENSE.md,sha256=aYSSIb-5AFPeITTvXm1UAoe0uYBiMmSS8flvXaaFUks,10174
563
+ mct_nightly-2.2.0.20241128.546.dist-info/METADATA,sha256=0CvdGOzW-TiaTXZdjW8IyWZflelwtWDxF7mH95b3H-0,26446
564
+ mct_nightly-2.2.0.20241128.546.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
565
+ mct_nightly-2.2.0.20241128.546.dist-info/top_level.txt,sha256=gsYA8juk0Z-ZmQRKULkb3JLGdOdz8jW_cMRjisn9ga4,26
566
+ mct_nightly-2.2.0.20241128.546.dist-info/RECORD,,
model_compression_toolkit/__init__.py CHANGED
@@ -27,4 +27,4 @@ from model_compression_toolkit import data_generation
27
27
  from model_compression_toolkit import pruning
28
28
  from model_compression_toolkit.trainable_infrastructure.keras.load_model import keras_load_quantized_model
29
29
 
30
- __version__ = "2.2.0.20241127.000529"
30
+ __version__ = "2.2.0.20241128.000546"