https://github.com/teddy-van-jerry/ptcq
PyTorch Complex Quantization
https://github.com/teddy-van-jerry/ptcq
complex python pytorch quantization
Last synced: 4 months ago
JSON representation
PyTorch Complex Quantization
- Host: GitHub
- URL: https://github.com/teddy-van-jerry/ptcq
- Owner: Teddy-van-Jerry
- License: mit
- Created: 2023-03-28T08:00:32.000Z (about 3 years ago)
- Default Branch: master
- Last Pushed: 2023-03-30T08:55:22.000Z (about 3 years ago)
- Last Synced: 2025-02-03T21:48:20.144Z (over 1 year ago)
- Topics: complex, python, pytorch, quantization
- Language: Python
- Homepage: https://ptcq.tvj.one
- Size: 33.2 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# PTCQ
PyTorch Complex Quantization
[](https://pypi.org/project/ptcq/)
[](LICENSE)
> **Note** The quantization is used for result simulation only.
> It is **not** intended for computation acceleration.
## Installation
Install with `pip`:
```
pip install ptcq
```
Install from GitHub Repo:
```
pip install git+https://github.com/Teddy-van-Jerry/ptcq.git
```
## Limitations
PTCQ is actually **fake** quantization that simulates the quantization behavior.
- It does not consider the internal computation quantization process;
- It does not accelerate the computation, still requiring `float` precision;
- Quantization operations have to be implemented manually.
## Implementations
So far, fixed-point quantization for real and complex input has been implemented.
View [API documentation](https://ptcq.tvj.one/en/latest/api/fixed/) for more information.
- `ptcq.FixedQ`: Fixed-point quantization.
- `ptcq.FixedCQ`: Fixed-point complex quantization.
## License
This package is distributed by an [MIT License](LICENSE).