| Name | mct-quantizers-nightly JSON |
| Version |
1.5.2.20241023.post1317
JSON |
| download |
| home_page | None |
| Summary | Infrastructure for support neural networks compression |
| upload_time | 2024-10-23 00:13:20 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.6 |
| license | None |
| keywords |
|
| VCS |
|
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# Model Compression Toolkit (MCT) Quantizers
The MCT Quantizers library is an open-source library developed by researchers and engineers working at Sony Semiconductor Israel.
It provides tools for easily representing a quantized neural network in both Keras and PyTorch. The library offers researchers, developers, and engineers a set of useful quantizers, along with a simple interface for implementing new custom quantizers.
## High level description:
The library's quantizers interface consists of two main components:
1. `QuantizationWrapper`: This object takes a layer with weights and a set of weight quantizers to infer a quantized layer.
2. `ActivationQuantizationHolder`: An object that holds an activation quantizer to be used during inference.
Users can set the quantizers and all the quantization information for each layer by initializing the weights_quantizer and activation_quantizer API.
Please note that the quantization wrapper and the quantizers are framework-specific.
<img src="https://github.com/sony/mct_quantizers/raw/main/quantization_infra.png" width="700">
## Quantizers:
The library provides the "Inferable Quantizer" interface for implementing new quantizers.
This interface is based on the [`BaseInferableQuantizer`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py) class, which allows the definition of quantizers used for emulating inference-time quantization.
On top of `BaseInferableQuantizer` the library defines a set of framework-specific quantizers for both weights and activations:
1. [Keras Quantizers](https://github.com/sony/mct_quantizers/tree/main/mct_quantizers/keras/quantizers)
2. [Pytorch Quantizers](https://github.com/sony/mct_quantizers/tree/main/mct_quantizers/pytorch/quantizers)
### The mark_quantizer Decorator
The [`@mark_quantizer`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py) decorator is used to assign each quantizer with static properties that define its task compatibility. Each quantizer class should be decorated with this decorator, which defines the following properties:
- [`QuantizationTarget`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py): An Enum that indicates whether the quantizer is intended for weights or activations quantization.
- [`QuantizationMethod`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/quant_info.py): A list of quantization methods (Uniform, Symmetric, etc.).
- `identifier`: A unique identifier for the quantizer class. This is a helper property that allows the creation of advanced quantizers for specific tasks.
## Getting Started
This section provides a quick guide to getting started. We begin with the installation process, either via source code or the pip server. Then, we provide a short example of usage.
### Installation
#### From PyPi - mct-quantizers package
To install the latest stable release of MCT Quantizer from PyPi, run the following command:
```
pip install mct-quantizers
```
If you prefer to use the nightly package (unstable version), you can install it with the following command:
```
pip install mct-quantizers-nightly
```
#### From Source
To work with the MCT Quantizers source code, follow these steps:
```
git clone https://github.com/sony/mct_quantizers.git
cd mct_quantizers
python setup.py install
```
### Requirements
To use MCT Quantizers, you need to have one of the supported frameworks, Tensorflow or PyTorch, installed.
For use with Tensorflow, please install the following package:
[tensorflow](https://www.tensorflow.org/install),
For use with PyTorch, please install the following package:
[torch](https://pytorch.org/)
You can also use the [requirements](https://github.com/sony/mct_quantizers/blob/main/requirements.txt) file to set up your environment.
## License
[Apache License 2.0](https://github.com/sony/mct_quantizers/blob/main/LICENSE.md).
Raw data
{
"_id": null,
"home_page": null,
"name": "mct-quantizers-nightly",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.6",
"maintainer_email": null,
"keywords": null,
"author": null,
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/30/2a/84a5ad870e9ffccb120606be28aa9d7c8698399dd0f32f3e6968c8446807/mct-quantizers-nightly-1.5.2.20241023.post1317.tar.gz",
"platform": null,
"description": "# Model Compression Toolkit (MCT) Quantizers\n\nThe MCT Quantizers library is an open-source library developed by researchers and engineers working at Sony Semiconductor Israel. \n\nIt provides tools for easily representing a quantized neural network in both Keras and PyTorch. The library offers researchers, developers, and engineers a set of useful quantizers, along with a simple interface for implementing new custom quantizers.\n\n## High level description:\n\nThe library's quantizers interface consists of two main components:\n\n1. `QuantizationWrapper`: This object takes a layer with weights and a set of weight quantizers to infer a quantized layer.\n2. `ActivationQuantizationHolder`: An object that holds an activation quantizer to be used during inference.\n\nUsers can set the quantizers and all the quantization information for each layer by initializing the weights_quantizer and activation_quantizer API.\n\nPlease note that the quantization wrapper and the quantizers are framework-specific.\n\n<img src=\"https://github.com/sony/mct_quantizers/raw/main/quantization_infra.png\" width=\"700\">\n\n## Quantizers:\n\nThe library provides the \"Inferable Quantizer\" interface for implementing new quantizers. \nThis interface is based on the [`BaseInferableQuantizer`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py) class, which allows the definition of quantizers used for emulating inference-time quantization.\n\nOn top of `BaseInferableQuantizer` the library defines a set of framework-specific quantizers for both weights and activations:\n1. [Keras Quantizers](https://github.com/sony/mct_quantizers/tree/main/mct_quantizers/keras/quantizers)\n2. [Pytorch Quantizers](https://github.com/sony/mct_quantizers/tree/main/mct_quantizers/pytorch/quantizers)\n\n### The mark_quantizer Decorator\n\nThe [`@mark_quantizer`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py) decorator is used to assign each quantizer with static properties that define its task compatibility. Each quantizer class should be decorated with this decorator, which defines the following properties:\n - [`QuantizationTarget`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/base_inferable_quantizer.py): An Enum that indicates whether the quantizer is intended for weights or activations quantization.\n - [`QuantizationMethod`](https://github.com/sony/mct_quantizers/blob/main/mct_quantizers/common/quant_info.py): A list of quantization methods (Uniform, Symmetric, etc.).\n - `identifier`: A unique identifier for the quantizer class. This is a helper property that allows the creation of advanced quantizers for specific tasks.\n\n## Getting Started\n\nThis section provides a quick guide to getting started. We begin with the installation process, either via source code or the pip server. Then, we provide a short example of usage.\n\n### Installation\n\n#### From PyPi - mct-quantizers package\nTo install the latest stable release of MCT Quantizer from PyPi, run the following command:\n```\npip install mct-quantizers\n```\n\nIf you prefer to use the nightly package (unstable version), you can install it with the following command:\n```\npip install mct-quantizers-nightly\n```\n\n#### From Source\nTo work with the MCT Quantizers source code, follow these steps:\n```\ngit clone https://github.com/sony/mct_quantizers.git\ncd mct_quantizers\npython setup.py install\n```\n\n### Requirements\n\nTo use MCT Quantizers, you need to have one of the supported frameworks, Tensorflow or PyTorch, installed.\n\nFor use with Tensorflow, please install the following package:\n[tensorflow](https://www.tensorflow.org/install),\n\nFor use with PyTorch, please install the following package:\n[torch](https://pytorch.org/)\n\nYou can also use the [requirements](https://github.com/sony/mct_quantizers/blob/main/requirements.txt) file to set up your environment.\n\n## License\n[Apache License 2.0](https://github.com/sony/mct_quantizers/blob/main/LICENSE.md).\n\n\n",
"bugtrack_url": null,
"license": null,
"summary": "Infrastructure for support neural networks compression",
"version": "1.5.2.20241023.post1317",
"project_urls": null,
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "a22965c6c716f2b4bc5bc2dac7080b582ae02502786398a74183a2fa0e5ad8b7",
"md5": "1fa29dfe41d06c19fdf8a4d7f1cc04a7",
"sha256": "aca29dfeca40dbd1d5271d67cd41f38c3a105b76553a917fe27c327b59e69504"
},
"downloads": -1,
"filename": "mct_quantizers_nightly-1.5.2.20241023.post1317-py3-none-any.whl",
"has_sig": false,
"md5_digest": "1fa29dfe41d06c19fdf8a4d7f1cc04a7",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.6",
"size": 105095,
"upload_time": "2024-10-23T00:13:18",
"upload_time_iso_8601": "2024-10-23T00:13:18.677656Z",
"url": "https://files.pythonhosted.org/packages/a2/29/65c6c716f2b4bc5bc2dac7080b582ae02502786398a74183a2fa0e5ad8b7/mct_quantizers_nightly-1.5.2.20241023.post1317-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "302a84a5ad870e9ffccb120606be28aa9d7c8698399dd0f32f3e6968c8446807",
"md5": "658b3ce2847de5039df31885bb35d574",
"sha256": "1cf4f35c6b7acd23115b11c07d21a187a20b3d2fffbd06182a9411ec65ab34b1"
},
"downloads": -1,
"filename": "mct-quantizers-nightly-1.5.2.20241023.post1317.tar.gz",
"has_sig": false,
"md5_digest": "658b3ce2847de5039df31885bb35d574",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.6",
"size": 47208,
"upload_time": "2024-10-23T00:13:20",
"upload_time_iso_8601": "2024-10-23T00:13:20.414249Z",
"url": "https://files.pythonhosted.org/packages/30/2a/84a5ad870e9ffccb120606be28aa9d7c8698399dd0f32f3e6968c8446807/mct-quantizers-nightly-1.5.2.20241023.post1317.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-10-23 00:13:20",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "mct-quantizers-nightly"
}
