quant-pack is a versatile and easy-to-extend Python package for neural
network quantization study.
Following original works are conducted with quant-pack.
| Work | Publication | Module / Sub-package in quant-pack |
|---|---|---|
| GQ-Nets | Li et al. Rejected by ICLR 2020 * | inverse_distillation |
| FQN for Detection | Li et al. CVPR 2019 | coming soon |
* We decided not to put this work on arXiv until we made enough money with this technique.
We reproduced following third-party works in quant-pack. Links to the
origin publication and related code module are listed.
| Work | Publication | Module / Sub-package in quant-pack |
|---|---|---|
| LR-Nets | Shayer et al. ICLR 2018 | variational |
| Feature Denoising | Xie et al. CVPR 2019 | NonLocal |
quant-pack is distributed under Anti-996 license (V1.0, draft). See LICENSE for details.