-
## 一言でいうと
組込機器などの小メモリ環境でNNを動かす研究。Binary NNは重みを0/1で持つことで省メモリ化したが、中間の計算結果は浮動小数点でありこれがメモリを大きく食っている。そこで畳み込みをシリアルに行っていくことでこれを解消した。BNNをEmbedded向きにした、名付けてeBNN。
**Repository:** https://github.com/umutcanaltin/fpgai_compiler
**Branch with paper.md** (empty if default branch): main
**Version:** v1.0.0
**Edit…
-
> [0] Bengio, Yoshua, Ian J. Goodfellow, and Aaron Courville. "Deep learning." An MIT Press book. (2015). [pdf] (Deep Learning Bible, you can read this book while reading following papers.) :star::sta…
-
Thx for your work again, i have tried your default config for w4a4g2 quantization. it works well for resnet-18 on imagenet(top1 acc ~71%). So i want to try if it can work for w1a1(a.k.a BNN). I use th…
-
I have 2 models one is baseline keras model and its equivalent keras model, the models are taken from the QKerasTutorial.ipynb,
My keras model is shown below:
```
Model: "model"
_________________…
-
Hi folks,
I noticed the discrepancy where:
``` bash
abc -c "read_truth -x 11111111; bdd; print_kmap; print_io"
```
throws an error saying:
```
Cannot create constant function.
Segmentation …